Dissertationchap3

Jeffrey G. Nicholson‘s Dissertation

CHAPTER III
CONCEPTUAL FRAMEWORK AND METHOD

Conceptual Framework
The conceptual framework of the study was based on Donabedian’s classic
structure/process/outcome (SPO) model as a tool for assessing health care quality (Donabedian,
1966; Burns, 1995). Donabedian defined structural measures of quality as the professional and
organizational resources associated with the provision of care, such as staff credentials and
facility operating capacities. Process measures of quality refer to the tasks done to and for the
patient by practitioners in the course of treatment (Gustafson & Hundt, 1995). Outcome
measures are the desired states resulting from care processes, which may include reduction in
morbidity and mortality, and improvement in the quality of life (Kane & Kane, 1988).
Practitioner safety is a factor in the process of the SPO model while patient safety is the desired
outcome. This is exemplified by an adaptation of the Donabedian SPO health quality assessment
model proposed for the Australian government’s national health care system (Sibthorpe, 2004)
(see Figure 1).

Donabedian (1988) noted a distinction between two types of outcomes. Technical
outcomes encompass the physical and functional aspects of care. Examples of technical
outcomes include the absence of post-surgical complications and the successful management of
hypertension and other chronic conditions. Interpersonal outcomes encompass dimensions of the
“art” of medicine. These include patient satisfaction with care and the influence of care on the
patient’s quality of life as perceived by the patient. Within Donabedian’s framework, these two
types of outcomes are interdependent, so that one cannot be considered in isolation from the
other in evaluating the quality of care.

Figure 1.Framework for Performance Assessment in Primary Health Care GOVERNMENT PRIMARY HEALTH CARE SERVICES AND PROGRAMS
I. II. III. IV. Australian
STEWARDSHIP ORGANIZATIONAL STRUCTURES PROCESSES OF CARE OUTCOMES National Health Performance Framework
Policy development -clear objectives

Financing and Funding
-+/-incentives

Implementation
– contracting
– reporting requirements

Workforce development

IT infrastructure
support

R &D

Physical facilities and equipment

Staffing, including deployment

Staff training and development

Human resources management

Service organization and management, including protocols 

Financial management
Information systems

Needs assessment Performance assessment

Sick care (including curative, rehabilitative, palliative)

Health promotion

Disease Prevention

Advocacy Community development

 

 

 

Levels of health risk behaviors in client populations

Levels of
clinical status measures in client populations

Levels ofsatisfaction with care in client populations

 

 

 

TIER I Health Status and Outcome

TIER II Determinants of Health

TIER III
Health System
Performance: Effective Appropriate Efficient
Responsive
Accessible
Safe Continuous Capable Sustainable

Further, Donabedian (1966) asserted that the three categories of quality measures,
structure, process and outcome, are not independent but are linked in an underlying framework.
Good structure should promote good process and good process in turn should promote good
1 Adapted from Beverly Sibthorpe, Australian Primary Health Care Research Institute, 2004
outcome (Donabedian, 1988). This provided a theoretical rationale for linking outcome with
structure.

The variables of this study on practitioner safety were linked to structure, process and
outcome elements of Donabedian’s model. The elements of the Donabedian model that were
linked to study variables are bolded in Figure 1. The independent variables, the medical
practitioners, comprise a key staffing and deployment component of the organizational structure.
Their performance is influenced and affected by the organizational structure through the amount
of staffing provided, staff training and development, human resource management, adherence to
service protocols and practitioner performance assessment. The practitioner variables are also
key components of the process of care as they comprise the corps of care provision. The study
dependent variables, markers of practitioner safety, were linked to the organizational structure as
they are affected by staff training and development, human resource management, practitioner
adherence to protocols and practitioner performance assessment. The outcomes of patient
satisfaction and the safety of care provision were linked to both independent and dependent
variables. Patient satisfaction is influenced not only by the type of provider but by the provider’s
characteristics as portrayed by each dependent variable. In the study’s linkage to the Donebedian
model, safety of care itself was measured by the dependent variables as markers of safety. Figure
2 provides a model specification chart demonstrating the relationship between elements of the
Donabedian model and the study variables.

Figure 2. Model Specification Chart

Donabedian Framework Category Donabedian Framework Factors Proxy Variables
Organizational Structure Staffing, including deployment Physicians, PAs, APNs; Average years of practice; Practitioner gender
Organizational Structure Human resources management Physicians, PAs, APNs
Organizational Structure Service organization and management, including protocols Physicians, PAs, APNs; Average years of practice; Practitioner gender
Organizational Structure Performance Assessment Malpractice incidence, malpractice payments, clinical privileges actions, professional society actions, Federal program exclusions, DEA actions
Processes of care Sick Care Physicians, PAs, APNs
Outcomes Levels of satisfaction with care in client population Malpractice incidence (patient driven); Malpractice amounts
Outcomes Overall Safety of Care Provision Malpractice incidence, malpractice payments, clinical privileges actions, professional society actions, Federal program exclusions, DEA actions

Health care consumers play a significant role in the reporting and documentation of
practitioner safety as poor outcomes and patient harm are often brought to the attention of
authorities by consumers themselves. Donabedian further studied the role of the consumer in
quality assurance in the 1990s (Donabedian, 1992). Donabedian contended that consumers have
three main roles in the assurance of quality in health care: (a) consumers can be contributors to
quality by ?helping to define it, evaluating it, and providing information which will allow others
to evaluate it; (b) consumers can be targets for quality assurance ?by conceiving of practitioners
and patients as jointly engaged in the production of care,? when they are used as a means to
regulate practitioner’s behavior; and (c) consumers can be reformers of health care by direct
participation at the patient care level by provision of support by the administration which
empowers consumers to have an effect on the systems of care, ?by influencing the “markets” of
health care provision and by political action. Quality assurance is defined as “an activity aimed
to elicit information about clinical performance, and based on that information, to readjust the
circumstances and processes of health care” (p. 246). Donabedian concluded that when
consumers are allowed to help practitioners, they can make a considerable contribution to
enhancing the quality of care.

Donabedian’s framework of consumer participation in the assessment of health care
quality was essential to this study. The data used to drive the study have their bases in consumer
actions. That is, concerns about the quality of one’s medical care or that of a loved one are often
initiated by the health care consumer. Consumers have a variety of mechanisms for making
concerns known. The first mechanism is to approach the entities whose purpose includes the
monitoring of health care quality and consumer protection. These entities include hospitals,
better business bureaus, state medical licensing boards, insurance agencies, professional
associations and federal and state government regulatory agencies. A second option available to
consumers is to take civil action through the courts. A third is to voice concerns of safety through
the news media.

Since 2000 six states have enacted legislation supporting the creation of a state patient
safety center. These entities include: the Florida Patient Safety Corporation; the Maryland
Patient Safety Center; the Betsy Lehman Center for Patient Safety and Medical Error Reduction
(Massachusetts); the New York Center for Patient Safety; the Oregon Patient Safety
Commission; and the Pennsylvania Patient Safety Authority (Rosenthal & Booth, 2004).

All six centers are designed to house and coordinate statewide patient safety activities.
Specifically, patient safety centers are charged with promoting patient safety through a variety of
activities, which vary by state but may include:
• Educating health care providers and patients regarding processes that may reduce future
occurrences of adverse events;
• Developing systems of near miss and/or adverse event data reporting, collection,
analysis, and dissemination to improve the quality of health care;
• Fostering the creation of safety cultures to identify and determine the causes of adverse
events and near misses;
• Informing consumers about patient safety issues;
• Serving as a clearinghouse for the development, evaluation, and dissemination of best
practices;
• Promoting ongoing collaboration between the public and private sectors and
• Coordinating state agency initiatives (Rosenthal & Booth, 2004)

This study utilized data reported by the entities whose role is consumer protection,
entities that encourage consumer participation. The study’s data came directly from state medical
licensing boards, hospitals, professional societies with formal peer review, the office of the
Inspector General of the U.S. Department of Health and Human Services, malpractice payers,
and the U.S. Drug Enforcement Agency (see Figure 3).

With Donabedian’s SPO model as the theoretical and conceptual basis of the study, the
study itself set out to update, build upon and expand the limited work of researchers in the 1990s
who examined malpractice of physician assistants. Brock (1998) and Cawley et al. (1998)
independently examined the malpractice experience of PAs using data from the National
Practitioner Data Bank (NPDB). At the time, the data revealed that PAs had a very low rate of
malpractice judgments. Brock asserted that this factor would actually lead providers to hire PAs
as a way to reduce the risk of malpractice liability. Brock used data published in 1996 to
determine that there were significant differences in malpractice experiences of PAs and
physicians. Brock (1998) found that for one claim paid for every 46.6 physicians there was one
for every 808.1 PAs. The findings of Cawley’s group were similar and have been outlined in
Chapter II (Cawley et al., 1998). This study set out to develop a better and current understanding
of the earlier research and conclusions drawn.

There are several shortcomings to the earlier 1998 work which necessitated the further
investigation of this research. The researchers utilized only a subset of the data available – they
examined only medical malpractice payments. Additionally, the dataset used in 1998 was limited
to the first six years of the existence of the NPDB. During the first years of the NPDB,
underreporting of PA malpractice and misconduct was likely due to agency reporting of PA
misconduct under the name of the supervising physician. The current research examined not only
medical malpractice payments, but a variety of other adverse actions taken against PAs, APNs,
and physicians. Those actions are contained in the NPDB records database as outlined in Figure
3 and constituted the variables that were studied.

The methodology of the study was an analysis of the independent variables between the
three provider types for comparisons, relationships and statistical significance. The methodology
examined 324,285 total entries of medical malpractice payments and adverse actions taken
against providers in a 17 year study period. As outlined above, these variables were linked to
Donabedian’s framework of health care quality assessment though the framework’s inclusion of
patient and provider safety outcomes.

Research Questions
The intent of this study was to determine whether the practice of medicine by physician
assistants is as safe as the practice of medicine by physicians? Specifically, research questions
for this study included: (a) Do PAs negate their cost effectiveness through the costs of
malpractice?; (b) Is the rate of malpractice for physician assistants at the same trajectory as that
of physicians and advanced practice nurses?; (c) Is the ratio of malpractice claims per provider
the same for physician assistants, advanced practice nurses and physicians?; and (d) Are the
reasons for disciplinary action against PAs and APNs the same as those for physicians?

Hypothesis
Based on the limited prior research, it would not be unreasonable to assume that
physician assistant medical practice is at least as safe as the medical practice of physicians.
However, enough time has passed for a meaningful exploration, reliable national data are now
available, and the PA profession has grown considerably in size and scope. Therefore, this study
assumed the null hypothesis. That is, there is no statistically significant difference between the
safety of physician assistant medical practice, advanced practice nurse medical practice, and
physician medical practice as determined by malpractice medical payments and actions taken
against a practitioner’s ability to practice. The null hypothesis was also applied to each
dependent variable and for the research sub-questions. That is, it was hypothesized that PAs do
not negate their cost effectiveness through malpractice payments, the rate of malpractice
payments is the same and the ratio of malpractice claims per provider is the same for PAs,
physicians and APNs. Finally, it was hypothesized that the reasons for disciplinary action are the
same for these three provider groups.

Data Source
Data utilized in the study are a subset of data available in the National Practitioner Data
Bank (NPDB). The NPDB is a repository of national data on the incidence and amount of
malpractice payments by health care providers and actions taken by regulatory bodies against
health care providers’ ability to practice in the interest of patient and public safety. The NPDB
was created by the 1986 congressional Health Care Quality Improvement Act, also known as
Title IV of Public Law 99-660. According to the NPDB Guidebook, the intent of Title IV of
Public Law 99-660 was to improve the quality of health care by encouraging state licensing
boards, hospitals and other health care entities, and professional societies to identify and
discipline those who engage in unprofessional behavior; and to restrict the ability of incompetent
physicians, dentists, and other health care practitioners to move from state to state without
disclosure or discovery of previous medical malpractice payment and adverse action history.
Adverse actions can involve licensure, clinical privileges, and professional society memberships
(NPDB Guidebook, 2007). The Health Quality Improvement Act of 1986 requires hospitals,
other health care entities, professional societies, medical malpractice payers and the Office of the
Inspector General to report malpractice payments, adverse licensure actions, professional review
actions, clinical privilege actions, exclusions for Medicare and Medicaid programs and Drug
Enforcement Agency Actions to the NPDB within 30 days of the activity. All of the above
reporting is required for physicians and dentists. All of the above are required reporting for PAs
and APNs except licensure actions, clinical privilege actions and professional society actions.
The law also requires hospitals to query the NPDB prior to the granting of hospital privileges for
any credentialed health care provider and every two years thereafter.

This research study was a secondary analysis of the publicly available data file of the
NPDB. Reporting of the malpractice and adverse actions by states and U.S. territories to the
NPDB is required by federal law, although some data in the databank are voluntarily reported.
The NPDB 2007 data contain information on disclosed reports of malpractice payments and
adverse actions of health care practitioners from September 1, 1990 through December 31, 2007.
The full NPDB data consist of 414,404 cases and dozens of variables, including information
about the characteristics of a variety of healthcare providers with medical malpractice payments
and practice-limiting actions, not just physicians, PAs and APNs. The categories of actions
which define medical misadventure reported by the NPDB include those outlined in Table 5. The
NPDB maintains a website, and the public data are available for downloading and analysis
(http://www.npdb-hipdb.hrsa.gov).

Sample
A 17 year selection of all data collected by the National Practitioner Databank was used,
from January 1, 1991 through December 31, 2007, to examine a variety of factors and trends in
medical misconduct between three groups of practitioners (NPDB, 2008). The first and current
years of data were not used because data is incomplete. The number of total data entries for
physicians, physician assistants and advanced practice nurses during the period of examination
was 324,285.

Demographic Data
Demographic data on the number of active practitioners in each of the three provider
groups came from the considered most reliable sources. The number of physicians came from the
American Medical Association master file as reported in the AMA annual publication Physician
Characteristics and Distribution in the US. Physician assistant demographic data came from the
American Academy of Physician Assistants, a national association that conducts annual surveys
of its members. Procuring reliable data on APNs was more difficult. There is no central or
national professional association that represents advanced practice nurses. APN is a term that
encompasses at least four different advanced practice nursing professional designations including
nurse midwife, nurse anesthetist, clinical nurse specialist, and nurse practitioner. There are
multiple certifying bodies for these designations and competing national professional
associations, none of which survey all designations. To compound the difficulty in obtaining
accurate demographic data, nurses in advanced practice often designate themselves in multiple
advanced practice categories. For example, a clinical nurse specialist may also consider
themselves a nurse practitioner and report themselves as both on surveys. This is reported as an
inherent problem in the only national survey that includes all advanced practice nursing
designations, the National Sample Survey of Registered Nurses (NSSRN) conducted by the
Health Resources and Services Administration of the U.S. Department of Health and Human
Services, Bureau of Health Professions. The NSSRN disclaimer reports in part that NSSRN
samples RNs who may also claim APN preparation, numbers may include many who are not
currently practicing in their specialty but who were once prepared and completed an APN
program earlier in their careers, and that respondents could be certified in multiple specialties by
multiple organizations (U.S. HRSA, 2004). Although the APN numbers are known to be inflated,
researchers recognize that there is no other or better national database containing APN
demographic information over time and so researchers continue to use the HRSA APN data.
Therefore this researcher has also chosen to use the HRSA APN demographic data for numbers
of APN providers.

Figure 3.2 National Practitioner Data Bank at a Glance

National Practitioner Data Bank at a Glance
The National Practitioner Data Bank was established under Title IV of Public Law 99-660, the Health Care Quality Improvement Act of 1986. NPDB is an information clearinghouse to collect and release information related to the professional competence and conduct of physicians, dentists, and other health care practitioners.
Who Reports?
  1. Medical malpractice payers
  2. Medical/Dental State Licensing Boards
  3. Hospitals and other health care entities
  4. Professional societies with formal peer review
  5. HHS Office of Inspector General
  6. US Drug Enforcement Administration
  7. Federal and State Government agencies
  8. Health plans
What Information is Available?
  1. Medical malpractice payments (all health care practitioners)
  2. Adverse actions – based on reasons relating to professional competency and conduct (primarily physicians/dentists)
    1. Licensing actions: revocation, suspension, censure, reprimand, probation, surrender, denial of an application for renewal of license and withdrawal of an application for renewal of license (reported as a voluntary surrender)
    2. Clinical privileges actions
    3. Professional society membership actions
  3. Medicare and Medicaid exclusions (all health care practitioners)
  4. US Drug Enforcement Administration actions (all health care practitioners)
Who Can Query?
  1. Hospitals
  2. Other health care entities with formal peer review
  3. Professional societies with formal peer review
  4. Boards of Medical/Dental Examiners and other health care practitioner State Licensing Boards _ Plaintiffs’ attorneys or plaintiffs representing themselves (limited)
  5. Health care practitioners (self-query)
  6. Researchers (statistical data only)

Adapted from the National Practitioner Data Bank (2006). Retrieved from http://www.npdbhipdb.
hrsa.gov/pubs/Data_Banks_at_a_Glance.

Health care providers in this study were selected and reclassified into three types: (a)
physicians including allopathic physicians (MDs), osteopathic physicians (DOs) and physician
interns/residents; (b) physician assistants; and (c) advanced practice nurses (APNs) which
include nurse anesthetists, nurse midwives, nurse practitioners, advanced practice nurses and
clinical nurse specialists (see Appendix B).

Method and Research Design
The NPDB Public Use Data File was downloaded from the NPDB website (www.npdbhipdb.
hrsa.gov/publicdata.html). Data from January 1, 1991 through December 31, 2007 was
extracted for analysis. Four report types were reclassified into adverse action reports employing
data with formats in use before and after 11/22/1999. Malpractice payments were examined
using data formats before and after 1/31/2004. Health care providers in the study were
reclassified into three types: (a) physicians, including allopathic physicians (MDs), osteopathic
physicians (DOs) and physician interns/residents; (b) PAs; and (c) APNs. All other practitioners
were excluded.

The identified data were used to determine the following: trends in malpractice incidence,
payment amount and adverse action incidence; ratios of medical malpractice payments by
provider type; and comparisons of PAs to physicians and APNs in all variables studied. Payment
averages, median of payment, total of payment and total amount of payment (provided as 1991
dollars for prior study comparisons and also adjusted for inflation to constant 2008 dollars).
Inflation adjusted amounts were calculated using inflation percent for each year with a formula
generated by the Consumer Price Indexes of U.S. Department of Labor. Other variables in this
study included adverse licensing or credentialing actions, professional society actions, age group,
time in practice, gender, state of license, and basis for action.

A chi-square test (also chi-squared or ?2 test) is any statistical hypothesis test in which
the test statistic has a chi-square distribution when the null hypothesis is true, or any in which the
probability distribution of the test statistic (assuming the null hypothesis is true) can be made to
approximate a chi-square distribution as closely as desired by making the sample size large
enough. Specifically, a chi-square test for independence evaluates statistically significant
associations between proportions for two or more groups in a data set (Wikipedia, 2008). The
proportions of the groups being compared may be different but statistically associated. For this
study associations are being tested between dependent variables for three groups in the dataset:
physicians, PAs, and APNs. The chi-square distribution is a family of probability density curves
defined by the number of degrees of freedom. The degrees of freedom depend on the number of
categories and is calculated as (number of rows-1) X (number of columns-1). For example, if
there is a 2×2 table, the degrees of freedom are calculated as (2-1) x (2-1) = 1.
The formula used to calculate: ?2 = ? ?
ExpectedValue
(Observed ExpectedValue) **2
Statistical analyses used in this study included chi-square analyses to explore associations
among the dependent variables including provider’s year of practice, state of license, number and
amounts of medical malpractice payment, and number of type of adverse action reports. The chisquare is a good choice since we are most often comparing three groups and looking for
statistically significant associations between these three groups in the data set.
One-way analysis of variance (ANOVA) is used for a continuous outcomes for >2
(unpaired) independent groups. It is used to test for a difference in the mean outcome level
between three or more independent groups. If we have a significant result from ANOVA, we
may be interested in testing which of the groups differ from each other (post hoc tests) by using a
selected method such as that of Tukey or Scheffe for multiple comparisons. An ANOVA result is

significant if the result of at least one pair is significant (in our case we will describe this as a
significant difference). The null hypothesis is rejected if a statistically significant difference is
found to exist. When there is an unequal size such as among Physician, PA, and APN data,
Scheffe’s method ANOVA is used because it is a better choice.
A one-way ANOVA method was used for pair-wise comparisons among three types of
healthcare providers: PAs and physicians; APNs and physicians; and PAs and APNs. The
significance level was set at p ? 0.05. SAS version 9.1.3 for Windows was used to analyze data
(SAS Institute, Carry, NC).

Table 5. Variables Studied
______________________________________________________________________________
Independent Variables Dependent Variables
______________________________________________________________________________
Physician Assistants  Total Number of Malpractice Payments
Physicians (MD, DO) Average Amount of Malpractice Payments
Advanced Practice Nurses Average Years of Practice
 Total Number of Adverse Events/Actions
 State and Medical Board Licensing Actions
 Clinical Privileges Actions
 Professional Society Membership Actions
 Practitioner Exclusions from Medicare and Medicaid
 Programs
 U.S. Drug Enforcement Agency (DEA) Actions
 Year of Adverse Action
 Basis for Action
 State of License
______________________________________________________________________________

The independent variables used in the study were reported by the NPDB as the field of
license. The independent variables were defined in Chapter 2. Field values from the database for
each category of clinician, the independent variables, are presented in Appendix A. The total
number of adverse events variable included 52 different types of actions taken against a clinician
or clinician’s license (see Appendix B). The variables of state and medical board licensing
actions, clinical privileges actions, professional society membership actions, practitioner
exclusions from state or federal programs, and U.S. DEA actions were all reported separately by
the NPDB as sub-fields of adverse actions (see Appendix B). The variable of malpractice
payments included payments made by insurers, state patient compensation funds, excess
judgment funds or other similar state funds. Payment amounts were analyzed unadjusted and
adjusted for inflation by the researcher to 1991 dollars. The basis for action variable contained
149 causes for action against a clinician or clinician’s ability to practice (see Appendix C).
The following was determined: trends in malpractice incidence and amounts, trends in
other defined adverse actions, ratios of medical malpractice payments and defined adverse
actions, ranking from most common to least common bases for actions, ranking of malpractice
and adverse action incidence by state and comparisons between physicians and PAs, physicians
and APNs, and APNs and PAs for all variables studied. For the malpractice variables, payment
averages, median of payment and total amount of payment was calculated. Dollar values were
adjusted for inflation by changing all payments to 1991 dollars using inflation percent for each
year with calculated formula adapted from consumer price indexes of U.S. Department of Labor.
1991 dollars were chosen so that direct comparisons could be made with the work of Brock and
Cawley et al.

Data Presentation
The data were presented in table, graph and chart formats. The following data
presentations were presented:
• A set of tables, comparing the thee provider groups in each of the variables for
the full 17 year period
• Tables and graphs of trends in adverse actions for each dependent variable by
year
• Ratios of adverse actions per provider group and adverse actions per provider
62
• Years in practice at time of adverse action and age grouping at time of adverse
action
• Tables for each of the dependent variables comparing PA to physician, PA to
APN and physician to APN
• Summary tables comparing provider groups in basis for action, malpractice
payments and adverse actions

Limitations
This study of PA practice and currently observed liability issues has limitations. As with
most studies, the research was confined to the available data. These data may not be
representative of all current malpractice or liability cases that involve PAs. The data used were
provided solely through federal reporting requirements. It is possible that some cases involving
malpractice or disciplinary actions of PAs have never been reported, were settled outside of the
courts or regulatory agencies, or were reflected in a supervising physician’s records versus that
of a PA.

Liability and Specialty Differences
No comparison of malpractice incidence across disciplines is fair without an
understanding of the liabilities undertaken by each discipline. While this study demonstrated
differences in malpractice incidence, payment amounts, and adverse action incidence between
PAs, APNs and physicians, the reader is cautioned and reminded that each of these medical
provider groups is comprised of a different compilation of medical practice specialties with a
subsequent difference in malpractice risk. The data set utilized did not allow for direct
comparisons across the three provider groups by specialty of practice. Only APN midwives and
anesthetists were reported separately and only because they are certified separately from other
APNs.

Role Differences
Additionally, physician assistants at their founding were designed to be dependent
practitioners, working alongside physicians as their assistants rather than as their substitutes.
Although PA practice has become more autonomous than its founders may have anticipated in
the 1960s, it is generally recognized that PAs are not expected to possess the full medical
knowledge base of physicians nor are they expected to manage the most complicated of patients
without assistance from a supervising physician. Likewise, licensing and regulatory agencies
recognize that APNs do not possess the same degree of training as physicians and therefore
require a collaborating physician for APNs in much the same manner as a supervising physician
is required for PAs.

The reader is therefore cautioned to bear in mind is that PAs and APNs may not as a
whole take on the same level of malpractice risk as physicians. PAs and APNs may not perform
some medical care that carries inherently greater risk to the patient and higher liability for the
clinician. Some examples of this include that PAs in orthopedic and cardiovascular surgery may
assist in surgeries but actually not perform the surgery itself. PAs working in obstetrics may
provide both pre and post-natal care of the mother, but may not actually provide childbirth
services. Thus we may expect physicians in these specialties to have a greater number of
malpractice claims than the PAs working under them in these specialties. Conversely, an APN
nurse midwife may have an equal or even greater malpractice risk that of an obstetrician
physician because they do deliver babies, often far from a medical facility. The ability to account
for variations in risk by task or is not the intent of this study.

Likewise, it is not the intent of this study to determine what that difference in risk is
between these provider groups. The study is not intended to determine, define or quantify the
differences in liability or malpractice risk between PAs and physicians or PAs and APNs. It is
solely intended to analyze available data and report the differences in actual malpractice
incidence, payments and other known outcome markers of safety over a 17 year period.

Autonomy Differences

In order to assess the inherent differences in malpractice risk and liability between
physicians, PAs and APNs, one would need to both quantify the differences in autonomy
between PAs, APNs and physicians and to account, compare and proportion the variety of
medical specialties of each provider group, each having its own inherent risk. These tasks are
complex and well beyond the scope of this study. The question of autonomy differences alone is
difficult to quantify because the level of autonomy of a PA or APN is determined by multiple
factors and may vary greatly not only from one specialty to another but from one employer,
employment setting or supervising physician to another. The amount of autonomy of a PA or
APN is largely determined by the provider’s own confidence and comfort with the level of care
being provided. Since these two practitioner types were founded on the principle of extending
physician care as much as possible, state regulations have been written broadly to allow
physician extenders to push their training, knowledge and skills to its limits. Physicians, rather
than envisioning their role as delegating minor tasks or acting as gatekeepers of physician
extender practice, have allowed mid-level practitioners to set their own limits of care within the
supervising physician’s practice specialty. State regulations state simply that PAs may not
practice outside the scope of their supervising physician’s board specialization. The PA or APN
approaches the supervising or collaborating physician for assistance on an as-needed basis.

Autonomy may also vary by employment setting or employer guidelines. For example,
some emergency room physician groups require their PAs to discuss or “staff” every patient seen
by the PA, while others more commonly prefer that the PA only come to the supervising
physician when questions in care arise. Some emergency physicians allow PAs to see any patient
in line for service without regard to patient acuity or level of care, while others restrict their PAs
to seeing only “minor” emergencies or “urgent care.” The difficulty in generalizing or in
quantifying the autonomy issue has been an obstacle to research in this area. While there is some
limited research on the tasks that PAs perform as compared to physicians, there is no literature
on the level of autonomy in performing those tasks or the inherent malpractice risk in performing
those tasks.

Additional Limitations and Limitations Summary
Some of the variables studied were reporting requirements in the dataset for physicians
but not PAs or APNs. Professional society membership exclusions was one such variable. These
variables have been excluded. Finally, a limitation of the study was the inability to differentiate
providers into their medical specialties. It compared all physician assistants to all APNs and to
all physicians. There may be a greater proportion of physicians or APNs who work in inherently
higher risk specialties than PAs. Two such specialties that are more popular among APNs than
PAs are anesthesia and midwifery. A more exact comparison would be made by comparing PAs
who work in a specific specialty with their APN and physician counterparts who work in that
same specialty. This information could not be derived from the dataset. It would make an
excellent topic of a future study. However, even if specialty comparisons are made, the varying
levels of autonomy and role delineation between APNs, PAs, and physicians must also be
addressed.

Chapter Summary
The conceptual framework of Donabedian was chosen because it is the most well
accepted and highly regarded model for studying health care quality. Donabedian’s model
provided a framework where safety was defined as a critical component of health care outcomes.
It was argued that practitioner safety is a key component of the process function of the model as
a determinant of patient health outcomes, and that all components of the model are
interdependent. Patient outcomes rely on the interplay and interdependency of structure and
process. Chapter III reiterated the rationale for the study and explained how the study builds
upon the very limited earlier work on practitioner safety conducted in the 1990s. The current
study is an updated exploration of the earlier work and is significant not only because it was
based on ten additional years of data, but also because it compared additional variables that the
earlier studies could not, providing richer analysis and opportunities for further research. The
research design was an analysis of variables using either chi square or ANOVA as appropriate to
determine whether statistically significant differences existed between safety outcome measures
of three medical provider groups. The NPDB data source, the sample, and rational for the
demographic data used were also explained in Chapter III. Limitations of the dataset, a
reiteration of the limitations of the study, and a reiteration of the purpose of the study was also
provided.

Chapter IV, the next section, presents the analysis of the data with careful consideration
to study limitations. Chapter V discusses and summarizes the findings, answers the study
questions, makes recommendations to educational leaders, health care policymakers, to the PA
profession and its training programs and suggests further research.