Are physician assistants/associates (PAs) cost-effective? If so, then to whom do the benefits accrue? What happens to the output of a clinical practice when a PA joins the medical or surgical staff? Furthermore, what are the outcomes in terms of access to patient care services, level of quality of care, practice revenues, and productivity? These and other fundamental questions about the use of new healthcare professionals have been at the heart of health services research spanning half a century. Answers to questions are emerging and the results are revealing about PA labor economics.
With the introduction of PAs and NPs in the 1960s and 1970s, one of the first questions raised was whether the financial incentives were adequate to employ them.1 Others theorized that PAs and NPs were functioning at the periphery of medical care, suggesting their roles were more complementary than substituting, and “the degree of role challenge is a factor determining physician acceptance.”2 One early observer of PAs argued that their optimal role was in rural health, where the “channel for upward mobility” might be more favorable and the “ethics of charging for services rendered by physician assistants” less onerous.3 This premise that employment could best occur at the margins was supported when a task inventory analysis and patient charges were revenue-positive in two rural practices each employing a PA. Two authors (as well as employers), each writing independently, concluded that the PA was a worthwhile employment investment.4,5 But the question as to what degree PAs were cost-effective remained unanswered. Health services researchers wanted to know how and where PAs could and should be employed.
The first hypothesis-driven study of PAs, their employment benefit, and annual productivity in general medical care was undertaken in 1974 by Jane Cassels Record.2 Record, an economist, was awarded a contract from the Bureau of Health Manpower to examine the effect of physician substitution in a large, vertically integrated prepaid group practice medical care delivery system on the West Coast.
A broadly represented panel of physicians, PAs, NPs, medical sociologists, economists, and social workers were convened to probe the perimeters of PA substitutability and quality assurance. The methodology for the project included assessing total annual output of medical services, those services exclusively treated by PAs (assigned their own panels of patients and daily appointment schedules, but administratively under physician supervision per state licensure policy), time of an outpatient visit, and the division of labor when a PA and physician jointly handled a patient encounter. The data spanned 2 years (1972-1974) and included reason for visit, presenting complaint, associated morbidities, diagnoses, types of visits (initial or continuing), imaging, laboratory, procedures, referrals, outcomes of care, characteristics of the patients seen, and type of experience of physicians and PAs.
Prospective and retrospective methodologies were incorporated. The study focused on the time and degree of activity of five PAs and 18 physicians in a general medicine clinic. Variables included office visits, telephone visits, peripheral activities, medication requests, laboratory results, medical record reviewing, and charting. Noncontact time, administrative duties, and regularly scheduled duties were included in the total annual productivity results. As a matter of policy, all patients, when offered the services of a PA, had the option to see a physician instead. Finally, in the longitudinal management of a condition, the patient would automatically be scheduled with a physician when returning for the third visit of the same complaint or illness. All this went into the report (Figure 1).
Record's 1975 report on PAs.
With respect to overhead costs, the investigation revealed no appreciable differences in use of resources for a patient visit by type of clinician when patient characteristics and diagnoses were held constant. In the realm of variable costs, the differential between PAs and physicians was in the basic salary and fringe benefits.
For physicians, the average salary with fringe benefits was $47,626; for PAs, the average was $15,164 (30% differential: 1972 dollars). Cost-sensitive areas explored included triage errors, no show and cancellation rates, unfilled slots, and supervision time (defined as 30 minutes each half-day to review cases with the PA). Additional time searching reference books, or hallway consulting was factored in by direct time-motion observation of all clinicians.
The results of the non-patient contact analyses were the same for the PA as the physician within 0.5%. From this 2-year intensive analysis, using multiple strategies to capture all the activity of an employed PA in general medical care, the percentage of total primary-care office visits that a PA alone can manage, except for general physician supervision, was about 70%. However, it was clear that nearly 83% of the primary services could be handled by PAs if specific physician help was given in 8% of the total visits. The number rose to 12% of the total when certain encounters were considered PA-appropriate visits.6 This 12% reference was Kaiser Permanente's policy at the time that select activities such as routine BP checks should be directed to the PA.
A series of analyses evaluating PA performance followed the Record report. The result of the subsequent economic studies was that PAs not only could but did provide 60% to 100% of the services traditionally performed by primary care physicians without consultation.7-11 In these and other studies the PA substitution effect of 83% found by Record and colleagues was cited often.6 That the findings were confined to primary care in one HMO was lost to most readers and instead the 83% substitution ratio took on legendary characteristics easily passed on because the PA community was small and communication was easy.
Although the verdict on PA cost-effectiveness rests on certain assumptions and compromises, the results had certain striking effects. In essence the large-scale longitudinal Jane Record study showed that PAs could improve a system's efficiency even given differences between physician and PA work weeks. In a 1981 book that summarized the growing body of PA economic research, Record noted that “PA efficiency was assessed quite conservatively.”12
Harold C. Sox, MD, an internist, editor emeritus of Annals of Internal Medicine, associate editor of the New England Journal of Medicine, and leader in the Institute of Medicine, was highly influential in US health policy. In examining the body of research on PAs, Sox concluded that a PA should be able to “provide the average office patient with primary care that compares very favorably with care given by the physicians.”13
Jump forward four decades from those early works and ask What is the body of evidence on the cost-effectiveness and productivity of PAs? Dozens of studies have been undertaken during this era, enough to warrant a systematic review of the evidence. In 2021, van den Brink and colleagues sifted through 50 years of literature about PA cost-effectiveness. The search produced 4,855 titles, and after deduplication, screening, reviewing, a total of 42 studies met rigid inclusion criteria.14
The studies were global in scope, spanning eight countries and three continents (North America, Europe, and Africa). Nine studies had a prospective design, and the others retrospective. Using the Cochrane Review protocol and rigorous inclusion criteria, the authors characterized the economic and productivity studies of PAs in terms of quality and outcomes of care.14
First off, all studies were screened for the Risk of Bias in Non-randomized Studies of Interventions (ROBIN-I) tool and scored along 8 criteria. Three studies scored as critically biased in how the results were reported (Figure 2). Overall, the majority of 42 studies scored well in measurement of outcomes and how the results were reported.
Risk of bias: PA cost-effectiveness studies
Although the number of PAs observed in each study varied, five studies were national in origin and assessed hundreds of PAs who were involved in the care of thousands of patients.
In terms of their roles, PAs were employed as a substitute for traditional physician services in 34 studies and in five studies, PAs were in a complementary role that improved the throughput of medical or surgical services. The quality of care delivered by a PA was comparable to a physician's care in 22 studies and in 13 studies the quality of care exceeded that of a physician. Both labor costs and resource costs were lower when PAs delivered the care compared with physicians as matched control. Many of the studies were of good methodological quality and results pointed in the same direction; PAs delivered at least the same quality of care as physicians and in a few instances better outcomes. On some occasions, the costs of the delivered care were lower when managed by a PA.
Sometimes this efficiency was due to the PA's reduced labor cost and sometimes because the PA was more effective as a producer of care and activity than was commonly seen.
A note about omitted studies in this systemic review: Government reports and the results of demonstration studies that produced important findings are significant documents on the broad range of PA economic activity. However, in this rigorous appraisal, such reports were excluded because they were not peer-reviewed. Also excluded were studies in which the PA and NP were assessed as one type of clinician and were not distinguished one from another. The authors of those combined provider studies were asked for delineated data and a few times the data were made available. The studies of combined clinicians are important medical workforce contributions and as there is convergence in thinking PAs and NPs are often interchangeable.15
However, in some instances differences do emerge between the two medical providers.16 The importance of separating outcome differences of care between PAs and NPs is that clinical managers may want to consider weighting one clinician over the other, depending on the needs of the operation and patient population.
The results on the economics of PA employment, whether in peer-reviewed biomedical journals or in government reports, presents a stronger understanding that efficient outcomes of care emerge when a PA is a part of the medical team. With the increased corporatization of medical practices (and with them the medical staff) the efficiency and financial benefit that PAs and APRNs bring to a clinical practice is a driving part of the equation.17-19
For scholars undertaking econometric studies of medical labor, the recommendation is to include the characteristics of the clinicians as variable. This often-omitted variable should include the number of clinicians under study, their degree of experience, sex, how patient satisfaction was assessed, and whether the quality of care met contemporary criteria.15 More granular PA and physician information is needed to understand what could be the influencing or confounding variables that affect the actual outcome. And finally, missing from the cited articles is the fourth goal in the quadruple aim of healthcare--taking care of healthcare professionals. Who is looking after the wellbeing of the provider and how is it being done?20