What is known about this topic
- • Physician assistants (PAs) function under doctor supervision.
- • The generalist training of PAs permits widespread use in most areas of medicine.
- • PAs are educated in the physician model of medicine.
Shortages of primary care doctors are occurring globally; one means of meeting this demand has been the use of physician assistants (PAs). Introduced in the United States in the late 1960s to address doctor shortages, the PA movement has grown to over 75 000 providers in 2011 and spread to Australia, Canada, Great Britain, the Netherlands, Germany, Ghana and South Africa. A purposeful literature review was undertaken to assess the contribution of PAs to primary care systems. Contemporary studies suggest that PAs can contribute to the successful attainment of primary care functions, particularly the provision of comprehensive care, accessibility and accountability. Employing PAs seems a reasonable strategy for providing primary care for diverse populations.
We undertook a review about physician assistants (PAs) in primary care – a foundation for understanding how they contribute to the provision of primary care. The literature is large enough to offer some generalisations, but a secondary goal is to alert the reader to significant gaps in current knowledge. For historical reasons, it is written with an American backdrop but with an eye on the global expansion of PAs and an international readership. Nurse practitioners (NPs) are mentioned when the literature reviewed combines PAs/NPs into a single provider category. However, Laurant et al. (2009) have extensively reviewed the combined literature on PAs and NPs in a variety of roles, and this work will not be repeated here. Instead, this undertaking focuses on useful and contemporary studies to guide employers and policy-makers as to whether PAs are suitable providers and how they may best be utilised in primary care.
Physician assistants are recognised as health professionals who practise medicine in collaboration with doctors through delegated clinical tasks and patient management (Cooper et al. 1998, Lowes 2000, Larsson & Zulkowski 2002, Hooker 2004, Parle et al. 2006). As of 2011, the United States has approximately 75 000 clinically active PAs and produces approximately 7000 graduates annually; there are almost 1000 PAs in other countries (with substantial growth predicted over the next decade) (Hooker 2010a). Their flexibility and generalist training permits them to function as providers under the supervision of a doctor in a variety of medical specialties and healthcare settings (American Academy of Physician Assistants [AAPA] 2009, Hooker et al. 2010b). One-third (34%) is employed in primary care settings (defined as family medicine, general medicine and general paediatrics). Because of the uniqueness of the American healthcare system, it is unclear whether the benefits experienced from the introduction of PAs as primary care providers will be experienced similarly in other healthcare systems (Hooker 2005). Nor has their use in primary care been consolidated in any systematic fashion that would be useful to potential employers. To address this gap, a review of PAs in primary care was undertaken. A description of PAs in primary care in the United States is provided and one question was posed: To what extent do PAs contribute to effective, safe and efficient team-based primary care?
Because of the nature of introducing a new health professional into a doctor-dominated system, policy analysts must assess the performance of PAs in relation to contemporary healthcare delivery. Healthcare organisations interested in evaluations that are not revenue based are likely to be interested in the effectiveness of PAs. At the heart of the question is identifying the services PAs can perform within the context of a physician–PA primary care team that will be beneficial to the practice, the patient, the employer and society.
The purpose of this review is to synthesise the available evidence regarding the contribution of PAs to primary care. While the PA profession has been in existence for over 40 years, the empirical study of this profession has been limited and definitive studies that comprehensively evaluate PA contributions to the functions of primary care are lacking. In the light of these limitations, a purposeful, policy-relevant review was conducted which categorised the studies based on a generally accepted theoretical model of primary care and relevance to the review question rather than evaluations of study methodology (Gough 2007).
The English language literature for all publications that include PAs in primary care was searched in these citation indices: Google Scholar, PubMed, CINAHL, Medline, Cochrane Database of Systematic Reviews, DARE, Embase, AHRQ, British Library Integrated Catalogue, ProQuest Dissertations and Theses, Sociological Abstracts, World Health Organization (WHO) and Web of Science. Search terms included ‘physician assistant’, ‘physician’s assistant,’‘Medex,’‘physician associate,’‘primary care,’‘family medicine’ and ‘general practice.’ Because PA development during the first two decades was small and the contributions to the literature of marginal value and relevance to contemporary primary care, the search was limited to 1990 through 2010. Studies were included in the review if an observational or experimental design was utilised, primary care physicians and PAs were compared relative to an outcome of interest, and studies were subjected to peer review. Post priori, the studies were sorted into the following areas of interest: description of PAs in primary care (distribution, skills, productivity and role) and contributions to the key functions of primary care (WHO 2008) (comprehensiveness, coordinated care, continuity of care, accessibility to care and patient centred), effectiveness, safe care and efficiency. The majority of the included studies were conducted in the United States, with 11 undertaken in four other countries. Studies excluded were grey literature, editorials, reworked original data and anecdotal findings.
A total of 93 papers and one monograph had comparisons of doctors and PAs in primary care. Primary care was mentioned in each of these papers but sometimes the use was broad and included emerging roles such as hospitalist, oncology, psychiatry, geriatrics or experiments in healthcare delivery such as telemedicine and retail medicine. When this search was refined, 42 papers and one monograph were considered useful to make comparisons between primary care doctors and PAs.
From the beginning of the concept of the PA, and before the term ‘primary care’ was widely used, the PA occupation was created to diagnose and treat common medical conditions in a general practice environment (Jones 2007a,b). It may have been this generalist background that led to their success. The emergence of the PA as an adjunct in delivering primary care services occurred in the early 1980s when the roles between family medicine doctors and PAs became less distinct and primary care more broadly defined (Bodenheimer & Pham 2010).
National studies of primary care visits to PAs, NPs and doctors in the United States suggest utility in the public sector, especially in the Department of Veterans Affairs and the Department of Defence, where ratios of PAs to doctors is high (Hooker 2008, Wozniak 1995). Similar ratios of PAs to doctors are not as common across the private sector, although some vertically integrated healthcare systems have staffing ratios of PAs to family practitioners as high as 50% (Hooker & Freeborn 1991). Nationally, the tendency to employ PAs in private practice family and general medicine settings are considerably less than in publicly funded community health centres (Hing et al. 2010). Reasons drawing PAs into non-primary care centres on higher remuneration in procedural specialties – surgery, dermatology and emergency medicine – lead the list (Morgan & Hooker 2010)
By the 1990s, evidence had emerged that PAs could expand the delivery of traditional physician-directed services (Osterweis & Garfinkel 1993, Schroeder 2002). Consequently, federal policies were enacted to ensure adequate reimbursement for PA services and to encourage the employment of PAs in rural underserved areas (Henry et al. 2011). Some of these primary care PAs serve rural and remote populations that could not attract a doctor (Henry & Hooker 2007).
As of 2011, approximately 75 000 PAs are considered clinically active across seven countries (Australia, Canada, the Netherlands, South Africa, UK, Ghana and the United States). In the United States, PAs/NPs provide at least 11% of all outpatient medical services. They work in every state and many domains of the federal government (Hooker 2008, AAPA 2009). All of the major medical and surgical specialties employ PAs and their role continues to expand (Morgan & Hooker 2010). The per cent of PAs in primary care has also been undergoing a shift since the mid-1990s owing to a number of influences. The most often cited reason is decreasing federal funding for PA education, that at one time emphasised primary care and deployment to underserved areas (Cawley 2008). Other influences include high-wage differentials for emergency medicine, dermatology, surgery and procedure-based roles, along with career dissatisfaction based on long hours, high stress, poor reimbursement and erosion of scope of practice (Phillips et al. 2009). Approximately 30–40% of clinically active PAs practise in primary care–related specialties (Hooker 2004, 2008, AAPA 2009).
All of the states, four US territories and four Canadian provinces enable PAs to practise; however, the ratio of PA to population is irregular, and the distribution is uneven (Sutton et al. 2010; Jones & Hooker 2011). Generally speaking, those states/provinces that have high concentrations of universities hosting PA programmes have high ratios of PAs, with the Northeastern states having the highest concentration of PAs, PA education programmes, and people than the rest of the United States (Liang 2010). Five universities in the Netherlands, four in Canada, three in the UK and one in Australia and Saudi Arabia also educate PAs (Hooker & Kuilman 2011).
The distribution of PAs also varies by metropolitan status. Physician assistants are more likely to work in urban settings than in rural ones; usually, the larger the metropolitan area, the greater the concentration of PAs. When non-metropolitan practices are examined, however, the proportion of PAs in rural practice is at least 9%; in some geographical regions such as the far west of the United States, the ratio of PAs/NPs to population is greater than doctors (Grumbach et al. 2003, Pedersen et al. 2008). The vast majority of rural practice PAs are in primary care (AAPA 2009, Henry et al. 2010). Geographical distribution of PAs in other countries is not known.
A national US practice analysis was undertaken in 2004 to assess the knowledge and clinical performance of PAs in practice. Such practice analysis is important to identify the range of skills and the set of beliefs about the domains of knowledge PAs need to possess to be competent in their careers. A total of 5282 completed surveys were considered representative of the PA population in years of experience, geographical distribution and practice specialty. The three skills required for most medical encounters were formulating the most likely diagnosis, basic science concepts and pharmaceutical therapeutics; it also revealed eight content domains. Overall, survey responses showed few differences in the tasks performed by PAs based on the length of time worked in the profession (Arbett et al. 2009).
Another study described the characteristics of providers, patients and the type of prescriptions written by PAs and NPs and compared these activities to those in metropolitan and non-metropolitan settings. A PA or NP was the provider of record for 3% of the primary care visits. The three providers wrote prescriptions for 60% of all visits, and the number of prescriptions was 1.3 per visit. PAs were more likely to prescribe a controlled substance than were physicians or NPs. In rural areas, NPs wrote more prescriptions than physicians and PAs, but both appear to prescribe in a manner similar to physicians in the type of medications used in their patient management (Hooker 2005). The majority of PAs in clinical practice appeared to be providing care in a manner similar to each other and similar to what ambulatory care doctors provide.
Analysis of productivity data from a national representative sample showed that, on average over 1 year, PAs performed 61.4 outpatient visits per week compared with 74.2 visits performed by physicians, for an overall physician full-time equivalent (FTE) estimate of 0.83. However, the productivity of PAs varied across practice specialty and location, with generalist PAs providing more visits than their specialist counterparts. Rural productivity was higher than urban, largely because of the concentration of generalist PAs in rural settings (Larson et al. 2001). Additionally, a policy analysis compared the productivity of solo practice physicians who employed PAs/NPs with those who did not, demonstrating that solo practice physicians who do employ PAs/NPs see an increase in the number of patient visits per week (127.2 vs. 116.4), a decrease in the number of weeks worked per year (47.7 vs. 48.6) and an increase in net income ($220 000 vs. $186 900), despite lower office visit fees ($90 vs. $96.50 for a new patient) (Wozniak 1995).
Additional productivity documentation can be found in a state-level analysis in Utah. Even though PAs make up only 6.3% of the state’s combined clinician (physician, PA, NP) workforce, they contribute approximately 7.2% of the patient care full-time equivalents in the state. The majority (73%) of Utah PAs work at least 36 hours per week and spend a greater percentage of total hours working in patient care when compared with physicians. The rural PA workforce reported working a greater number of total hours and patient care hours when compared with the overall statewide PA workforce (Pedersen et al. 2008).
In a Dutch study of a family practice doctor and an American-trained PA, the productivity, based on contacts per 1000 patients, increased by 17% over 1 year after the PA was added to a solo practice office. Measured per FTE of a GP, the number of GP contacts decreased slightly (2.3%). Types of contacts, diagnoses, drug prescriptions and new referrals to primary care of the GPs changed significantly. The number of PA contacts per 1 FTE PA was about 60% of that of the GPs, with clinical activities overlapping substantially. In the aggregate, the PA saw more women, children and patients aged 25–44 years, performed more practice consultations, made more women’s health-related diagnoses and prescribed more drugs for dermatological and respiratory problems (Simkens et al. 2009).
Physician assistants practice collaboratively with physicians to address the health needs of the population served. The role performed by any individual PA is negotiated with the supervising doctor and reflects the experience, training and preferences of all providers on the team, the needs of the patient population and the level of trust the physician has with the PA (Jacobson et al. 1998). As a result, there is the potential for significant overlap in the competencies of PAs and doctors. Five cross-sectional studies have compared visits performed by primary care doctors to PAs and NPs using a federal data set, which demonstrated that the main reason for a visit and the characteristics of patients seen by doctors and PAs/NPs were similar (Mills et al. 1998, Aparasu & Hegge 2001, Lin et al. 2002, Mills & McSweeney 2002, Genova 2006).
The clinical roles that PAs can perform within the provider team context fall under two broad categories (substitute and complement) and impact the distribution of patient care among team members (Jacobson et al. 1998, Starfield 1998). In both categories of labour, PAs are providing some form of substitution. When PAs perform all functions of primary care in a manner similar to physicians, they perform substitute roles. According to the philosophical ideal for primary care, this translates into assuming the tangible (the provision of the full complement of primary care services) as well as intangible (patient–provider relationship) attributes of a physician’s professional role for a subset of the patients (i.e. serve as a usual provider of care). One study examined the ability of practice attributes to predict PA performance in a substitute role and found that the most significant correlates included years in practice as a PA, years in practice with the supervising physician, annual income from the practice, practice in a rural county, recognition as the exclusive provider of primary care to patients and employment in a single-specialty group practice (Chumbler et al. 2001).
Physician assistants can also perform complementary roles in which they substitute for physicians for particular tasks. For example, they may be responsible for providing acute or preventive care to the population served by the provider team. Several studies have highlighted this type of labour for PAs and NPs in the provision of preventive care. A qualitative study that interviewed providers and administrators at nine large healthcare organisations examined the role of NPs and PAs. The researchers found that organisations perceived an important role of NPs and PAs in the delivery of preventive care to women owing to women’s preference for female providers, the shortage of female doctors and the tendency for PAs/NPs to place a higher focus on prevention (Coulter et al. 2000). Another study surveyed 1363 doctors about cancer screening in primary care patients and demonstrated that the majority of those surveyed (73–79% of family physicians and 60–70% of internists) are amenable to PAs providing such examinations. Of these, 631 physicians (46%) reported a PAs/NP performing at least one type of cancer screening examination on their patients, with family physicians more likely than general internists to use PAs/NPs to perform cancer screening examinations. Some evidence suggests that PAs and NPs may perform better at prevention tasks. One retrospective cohort study of 472 patient records that represented 16 million preventive healthcare visits among women aged 50–69 years was conducted. The relative risk ratios for breast examination and mammography during preventive visits across provider specialty and training types were compared. Across training degree types, PAs and NPs in primary care are more likely than medical doctors to adhere to cancer screening guidelines (Wallace et al. 2006).
Evidence suggests that PAs can substitute for physicians on a wide range of patient care tasks, supporting the contention that PAs have significant role flexibility (Hooker 2010a, Morgan & Hooker 2010). However, few empirical studies have evaluated the relationship between the role of the PA within the provider team and outcomes (Richardson et al. 1998, Scheffler 2008).
Commonly used measures of comprehensiveness include the scope of services provided and the rate of referral. Multiple studies have compared the scope of patient care services provided by PAs and physicians in primary care settings and have concluded that PAs can perform 85–90% of services traditionally provided by primary care physicians (Hooker 2010a). Ongoing national surveys of ambulatory medical care delivery systems demonstrate that PAs perform similarly to doctors within visits when types of patients are compared by primary diagnoses. In a study of 1200 US Community Health Centers, a higher percentage of PA visits were because of an acute condition (48%) compared with physician (34%) and NP visits (33%). Acute conditions were typically injury and illness. Patients with a comorbid chronic condition made up nearly half of all visits. The most frequent chronic conditions reported were hypertension, hyperlipidaemia, diabetes, depression, obesity, arthritis, asthma and chronic and obstructive pulmonary disease. The per cent of visits made by patients with any of these specific conditions did not vary by the type of clinician. Nor were there differences in the percentages of established patients seen by each type of clinician (87–89%). In these federally funded community health centres, which almost exclusively provide primary care, the staffing ratio of PA/NP to doctor averages 30% (range: 0–40%) (Hing et al. 2010).
Results from a study in Iowa suggest that comprehensiveness of primary care services varies by geographic location. Findings indicated that rural primary care providers performed more procedures than their metropolitan counterparts. Among 55 responding PAs, all reported patient education, prescribing, interpreting radiographs, referring patients and providing a wide range of services similar to their physician counterparts. Few differences emerged when comparing family medicine doctors with PAs in rural areas, suggesting that both clinicians are providing a broad array of medical services (Dehn & Hooker 1999).
Five studies on referral rates and patterns by PAs in primary care indicate that referring is an activity that does not substantially differ between PAs and doctors (Enns et al. 2003, Hooker 2004, Rubenstein et al. 2007, Kimball & Rothwell 2008, Simkens et al. 2009).
Care is coordinated when patients receive appropriate care in a cost-effective manner (Scheffler et al. 1996). Many conceptualisations of care coordination exist, but all agree that communication between primary care practitioners, other healthcare professionals and patients is a key component of coordination (Starfield 1998, Stille et al. 2005, Bodenheimer & Pham 2010). Coordination of care is generally viewed by primary care PAs as a function that falls within their clinical role (Jacobson et al. 1998). Specialist physicians report willingness to accept patient referrals from primary care PAs and general satisfaction with the appropriateness and timeliness of the referrals (Enns et al. 2003, Hooker 2004, Rubenstein et al. 2007, Kimball & Rothwell 2008, Simkens et al. 2009).
Continuity of care can refer to the transfer of information between episodes of care (informational continuity) or the provision of care over time by consistent providers (longitudinally and/or relational continuity) (Donaldson et al. 1996c, Haggerty et al. 2003, Cabana & Jee 2004). A patient–clinician relationship is a central feature of primary care; the potential for decreased relationship exists when a provider team approach is implemented. Two studies have evaluated the relationship between continuity and quality of care using a cross-sectional analysis of patient surveys. One evaluated the effects of visit continuity for patients (N = 14 835) of a large multispecialty practice served by primary care provider teams with PAs or NPs and the patient perceptions of the quality of primary care. Patients who only saw their primary care physician reported significantly higher physician–patient relationship quality and better assessments of organisational features of care (such as access and integration of care) than visits with providers other than their primary care physician. However, patients who had visits only with providers on their primary care team had significantly higher assessments of the clinical team, but lower assessments of their physician’s knowledge of them as a person than did those who had visits with providers off the team. The subgroup of patients that experienced visits with their primary care PA or NP team members reported better primary care experiences (Rodriguez et al. 2007).
Another survey of attendees of primary care clinics at five Department of Veterans Affairs medical centres (N = 21 689) evaluated the extent to which self-reported continuity of care related to patient satisfaction after adjusting for patient, provider and clinic characteristics. The mean adjusted humanistic score for patients who reported always seeing the same provider was 17.3 points higher than for those who rarely saw the same provider. Similarly, the mean adjusted organisational score was 16.3 points higher for patients who always saw the same provider compared with those who rarely saw the same provider. Demographic factors, socioeconomic status, health status, clinic site and patient utilisation of services were all associated with both the adjusted humanistic and organisational scores of the scale. There were no differences in type of provider (PAs/NPs or doctors) when distinguished by the patient, suggesting it was continuity of care, and not necessarily the type of provider, that was associated with higher patient satisfaction (Fan et al. 2005).
Accessible care is care that is easy for patients to obtain in a timely fashion (Donaldson et al. 1996a). Empirical evidence suggests that PAs can improve access to care to underserved patients and open access practices. The primary care patients of PAs, rather than doctors, are slightly more likely to be women, rural, uninsured or publicly insured. One study utilised administrative data and surveyed primary care clinicians including doctors, NPs, PAs and midwives in California and Washington to determine whether practice in underserved areas varied by provider type. PAs demonstrated a greater proclivity for providing care to the underserved as they ranked first or second in both states as the providers with the highest proportion of members practicing in rural areas, health professional shortage areas and vulnerable population areas (Grumbach et al. 2003). The finding that PAs practise in greater proportion than physicians and nurses in areas of low population density (i.e. rural areas) has also been identified in studies in Iowa and Utah (Dehn & Hooker 1999, Pedersen et al. 2008).
When compared with patients reporting primary care doctors as a usual source of care, patients of PAs were more likely to live in rural areas, lack insurance or have public insurance other than Medicare, report lower perceived access to care and/or have decreased likelihood of having some preventive care such as comprehensive health examination or mammograms. Despite these differences in characteristics and utilisation, there were no differences in patient complexity or in self-rated health between primary care patients of physicians and PAs, suggesting PAs can provide access to a usual source of care for a broad range of patients (Everett et al. 2009).
Appointment delays impede access to primary healthcare, and open access (OA) scheduling may improve the quality of primary health-care. A study assessed whether implementing OA during a 12-month period impacted practice and patient outcomes and differed by provider type. Providers (doctors, PAs and NPs) in four practices successfully implemented OA. On average, providers reduced their delay to the next available preventive care appointment from 36 to 4 days. No-show rates declined and overall patient satisfaction improved 16%. Continuity of care followed a similar pattern of improvement. Staff satisfaction neither improved nor declined. No significant differences in outcomes were seen by provider type, suggesting that PAs and doctors are similar in their adaptability to complex organisational changes aimed at improving access (Bundy et al. 2005).
Patient-centred care is recognised as a critical function of primary care, but agreement of the definition of this function is lacking. Most of the studies that have evaluated patient-centred care include patient satisfaction as an outcome (Mead & Bower 2002). Satisfying care, in this regard, means the patient completes the visit feeling their needs were met. No amount of quality care by the PA will overcome the stigma of unsatisfactory care if that is the way the patient perceives it.
To assess the extent to which the experiences of patients vary according to the type of primary care provider (PA, NP or doctor), a national, cross-sectional survey of the elderly patients receiving US government health insurance (Medicare) was undertaken. The beneficiaries completing the survey identified a primary care provider and recorded satisfaction data, patient socio-demographic characteristics, healthcare experience, types of care and types of supplemental insurance. A total of 146 880 completed surveys were analysed. While a small number (3770 or 2.8%) of respondents identified a PA or an NP as their sole personal provider, for questions on satisfaction with their personal care clinician, results were similar across the three providers. Patients who reported a physician as their primary care provider were more likely to have supplemental insurance as compared with patients who reported receiving care from a PA or NP. The conclusion was that Medicare beneficiaries are generally satisfied with their medical care and do not distinguish preferences based on the type of provider. For this group, the patient viewed all clinicians in primary care similarly across all patient characteristics (Hooker et al. 2005).
Effectiveness of primary care delivery may depend, at least in part, on using the correct mix of personnel. Starfield (1993) showed that the division of labour and economy of scale maximises the clinical capabilities of healthcare professionals. In primary care practice, it is neither necessary nor particularly efficient for each patient to be seen by a physician. As PAs are, by definition, physician-supervised clinicians, the very nature of their clinical role is to work with doctors in collaborative provider teams. To be effective, the PA needs to provide quality care to similar patients for similar diagnoses that result in outcomes comparable with those of a doctor. Several studies have been conducted which compare the care provided by PAs and doctors on quality measures including processes of care and/or patient outcomes for specific diagnoses.
In an effort to improve the cost-effectiveness of primary care for low back pain, an inclinic education intervention programme was designed to provide family practice doctors and PAs with specific information, tools and techniques that the literature suggested should be associated with more satisfying and cost-effective care. The back pain-related beliefs, attitudes and behaviours of 15 primary care providers in a large health maintenance organisation (HMO) clinic and of 14 family physicians in six group practices were assessed before and after the intervention. Significant increases were noted in the proportions of providers who felt confident they knew how to manage low back pain, who believed their patients were satisfied and who claimed they reassured patients that they did not have serious disease (whether they were PAs or doctors) (Bush et al. 1993).
PAs and NPs are primary care providers for patients with HIV in some clinics, but little is known about the quality of care they provide. A cross-sectional analysis compared eight quality-of-care measures assessed by medical record review. The quality of care provided by PAs/NPs was compared with that provided by physicians in 68 HIV care sites. The authors surveyed 243 clinicians (177 physicians and 66 NPs/PAs) and reviewed medical records of 6651 persons with HIV or AIDS. After adjustments for patient characteristics, most of the quality measures did not differ between NPs and PAs (and did not differ when compared with infectious disease specialists or generalist HIV experts). Adjusted rates of purified protein derivative testing and Papanicolaou cervical screens were higher for NPs and PAs (0.63 and 0.71, respectively) than for infectious disease specialists (0.53) or generalist HIV experts (0.47). PAs and NPs had higher performance scores than generalist non–HIV experts on six of the eight quality measures. The authors concluded that for the measures examined, the quality of HIV care provided by PAs/NPs was similar to that of physician HIV experts and generally better than physicians (non–HIV experts). Preconditions for this level of performance included high levels of experience, focus on a single condition, participation in teams and easy access to clinicians with HIV expertise (Wilson et al. 2005).
Three studies have evaluated the relationship between the type of provider and the attainment of treatment goals for diabetes, dyslipidaemia and hypertension. One cross-sectional analysis of 19 660 patients with diabetes, coronary artery disease or hypertension was conducted in the VA Connecticut Health Care System. While significant differences were seen in the type of patients cared for by PAs/NPs and resident physicians, the attainment of goals for each condition was similar, with one exception; PAs/NPs were more likely than resident physicians to attain a haemoglobin A1c (HgbA1c) goal of <7.5 (Federman et al. 2005). Another cross-sectional study of 46 family medicine practices measured adherence to American Diabetes Association guidelines via chart audits of 846 patients with diabetes. Compared with practices employing PAs, practices employing NPs were more likely to measure HgbA1c levels (66% vs. 33%), lipid levels (80% vs. 58%) and urinary microalbumin levels (32% vs. 6%) and to have treated for high lipid levels (77% vs. 56%). Practices with NPs were more likely than physician-only practices to assess HgbA1c levels (66% vs. 49%) and lipid levels (80% vs. 68%). However, these process improvements did not translate into improved outcomes, with the exception of better attainment of lipid targets in practices employing NPs. These effects could not be attributed to the use of diabetes registries, health risk assessments, nurses for counselling or patient reminder systems. Those practices employing either PAs or NPs were perceived as busier and had larger total staffs than physician-only practices. With regard to diabetes process measures in this study, family practices employing NPs performed better than those with physicians only and with those employing PAs (Ohman-Strickland et al. 2008). Similarly, a cohort study conducted on 88 682 primary care patients in 198 Veterans Administration clinics demonstrated that clinics that included NPs were associated with lower HbA1cs (approximately 0.31 percentage points) and clinics with PAs did not show a statistically significant difference in HbA1c when compared with clinics without PAs or NPs (Jackson 2011).
A randomised trial assessed the impact of a PA case manager responsible for screening, case finding and referral of geriatric primary care patients for common geriatric conditions (n = 792). Despite finding no difference in functional outcomes or hospital utilisation, patients who were provided with a PA case manager were more likely to have the target conditions identified (depression, cognitive and functional impairment, falls and urinary incontinence) than patients receiving standard care (Rubenstein et al. 2007). The findings suggest that incorporating PAs in supplemental roles for target populations can increase case finding, assessment and referral for previously underdiagnosed and treated conditions.
A study compared PA, NP and physician knowledge of the pharmacological management of Parkinson’s disease (PD) and their preferences for referring PD patients to specialists. PAs/NPs answered 46% and physicians answered 50% of questions about PD pharmacotherapy, in agreement with recommended clinical practice (P = 0.14). None of the providers differed in their preference to refer a PD patient to a specialist for management, but PAs and NPs were more likely to refer a patient to a specialist for diagnostic confirmation. The authors concluded that given trends for more PA and NP autonomy in patient care, it was reassuring that all providers had similar knowledge of PD pharmacotherapy. They also added that policies to substitute PAs and NPs for physicians might increase referrals to specialty providers for diagnostic confirmation (Swarztrauber & Graf 2007).
The results of the inputs and throughputs of care are critical, but if outcomes are unfavourable, the PA will be viewed as less effective than the doctor. The examination of differences in liability among professions is one way to assess the safety of care provided by healthcare professionals. A study assessing whether PA and NP utilisation increased liability analysed the malpractice incidence, payment amount and other measures of liability among doctors, PAs and advanced practice nurses (APNs). From 1991 to 2007, 324 285 entries were logged involving 273 693 providers of interest. Significant differences in liability reports were found among doctors, PAs and APNs. Physicians made, on average, malpractice payments twice that of PAs, but less than that of APNs. The probability of making a malpractice payment was 12 times less for PAs and 24 times less for APNs than physicians during the study period. For all three providers, missed diagnosis was the leading reason for a malpractice report; female providers incurred higher payments than men. Trend analysis suggests that the rate of malpractice payments for physicians, PAs and APNs has been steady and consistent with the growth in the number of providers. There were no observations or trends to suggest that PAs and APNs increase liability. From a policy standpoint, it appears that the incorporation of PAs and APNs into American society has been a safe undertaking, at least when compared with doctors (Hooker et al. 2009).
One analysis focused on whether PAs were cost-beneficial to employers. PAs and physicians within a large HMO saw four common acute medical conditions over 1 year. An episode approach was undertaken to identify all institutional resources used for a condition, and 12 700 medical office visits were analysed for each type of provider. Patient characteristics were controlled for age, gender and health status. A multivariate analysis identified significant cost differences in each cohort of patients. In every condition managed by PAs, the total cost of the visit was less than that of a physician in the same department. In no instance was a PA statistically different from physicians in use of laboratory and imaging costs and, in each instance, the total cost of the episode was less when treated by a PA. In some instances, PAs ordered fewer laboratory tests than physicians for the same episode of care. There were no differences in the rate of return visits for a diagnosis. When the type of provider encounters were further delineated by departments of family medicine, general internal medicine and paediatrics, the results remained the same. These findings suggest that PAs are not only cost-effective from a labour standpoint but also cost-beneficial to employers. In most cases, they order resources for diagnoses and treatment in a manner similar to physicians for an episode of care, but the cost of an episode of an illness is more economical overall when the PA delivers the care, which can be explained in part by the PA’s lower salary (Hooker 2002).
To estimate the savings in labour costs that might be realised per primary care visit from increased use of PAs and NPs in primary care, the practices of another HMO were examined; 26 primary care practices and data on approximately two million visits delivered by 206 practitioners were extracted from computerised visit records. On average, PAs/NPs provided one in three adult medicine visits and one in five paediatric medicine visits. Likelihood of a PA/NP visit was significantly higher than average among patients presenting with minor acute illness (e.g. acute pharyngitis). In adult medicine, the likelihood of a PA/NP visit was lower than average among older patients. Practitioner labour costs per visit (and total labour costs per visit) were lower among practice arrangements with greater use of PAs/NPs, standardised for casemix. The authors concluded that primary care practices that used more PAs/NPs in care delivery realised lower practitioner costs per visit than practices that used fewer PAs/NPs (Roblin et al. 2004, Roblin et al. 2011).
The literature regarding PAs in team-based primary care, spanning 1990 through 2010, demonstrates that these providers have enhanced certain aspects of the delivery of primary care. One of the attributes is that PAs have significant role flexibility, allowing healthcare systems multiple options for incorporating PAs into primary care provider teams. Evidence to date suggests that PAs can make significant contributions to select functions of primary care. These clinicians have also demonstrated inclination to provide care to underserved populations, thereby improving access to primary care. Available evidence suggests that the care provided by PAs is safe, effective and satisfying to patients insofar as it is comparable to doctors. Provision of care by PAs has a favourable cost benefit, thereby improving efficiency. The strength of this undertaking is that it identifies work where PAs are both cost-effective and complementary with primary care doctors in attaining the functions of primary care.
This study identifies where and how PAs can be utilised in primary care and at the same time demonstrates that the literature is far from complete. While the research conducted on PAs in primary care suggests their utilisation is favourable, all included studies had less than optimal scope and sample, limiting our capacity to make definitive statements about PAs in primary care. Significant work remains to be completed before we can claim to have a reasonable understanding of the scope of PA contributions to primary care.
Three broad areas of research are critical to optimising the contribution of PAs to primary care: role delineation, team processes and patient outcomes. Current theories regarding the roles of PAs in primary care have face validity, but the impact of these roles will not be understood until the theories are operational and empirically studied. Qualitative studies on team member perceptions of PA roles could assist in this endeavour. Understanding how roles impact team processes such as the provision of coordinated care and communication between team members is also critical. The literature on PAs in primary care is lopsided towards the United States, thus inhibiting international generalisations. Finally, evaluations of the impact of PA care on patient outcomes, particularly for chronic illness, are essential to understanding the full capacity of PAs to contribute to the delivery of team-based primary care.
Primary care is a bedrock principle in meeting the needs of society by providing integrated, accessible and accountable care. Current evidence suggests that the PA has surfaced as a valuable contributor to this important mission and is well suited to the provision of integrated care within provider teams in a variety of settings. The generalist training of the profession allows for overlapping competency with primary care doctors and has been a critical aspect of its adaptability. Access to safe and effective care is enhanced when PAs are part of the primary care provider team, and some patients will preferentially select them. These observations stand up to scrutiny and suggest that PAs are cost-effective to employers and probably cost-beneficial to institutions. However, limited empirical research has been undertaken comparing the process and the patient outcomes of coordination when performed by primary care PAs–doctor teams.
As the maximum substitution model of incorporating PAs in primary care was proposed by Record (1981), a number of policies have been enacted that not only permit the PA to work but to thrive. These primary care policies are being played out in a growing number of countries (Ashton et al. 2007; Mullan & Frehywot 2007; Jolly 2008, Farmer et al. 2009; Simkens et al. 2009). As demand for care rises and the supply of doctors fails to keep pace, the need for more primary care PAs will increase. Sustainability of any medical system involves organisational challenges and novel solutions. PAs may be one tool for meeting those challenges.
CE received financial support from AHRQ National Research Service Award (T32 HS00083); Community-Academic Partnerships core of the University of Wisconsin Institute for Clinical and Translational Research (UL1RR025011); and Health Innovation Program.