Table 1

Characteristics of studies included in full—studies presenting comparisons of PAs with other healthcare professionals

Specialty	Aim(s)	Study setting	Intervention	Comparison	Participants	Study design	Outcome measures	First author and year
Emergency medicine	To determine whether PAs are an appropriate option for providing services rendered by physicians in the ED	USA Walk in urgent care facility (satellite of an inner-city teaching hospital level 1 trauma centre)	PAs (n=5) rotate through the ED. PAs work solo from 08:00 to 12:00. No written diagnostic or therapeutic guidelines were followed.	25 physicians rotate through the ED. Physicians work solo from 17:00 to 21:00. No written diagnostic or therapeutic guidelines were followed.	n=5345 (seen by PAs) n=4256 (seen by physicians) during times of single coverage June 1995 to June 1996	Comparative retrospective	Length of visit Total charge	Arnopolin (2000)32
Emergency medicine	To examine the impact of PAs and nurse practitioners in EDs	Canada Six community hospitals with ED volumes between 23 and 66 000	PAs were introduced as an unregulated provider without medical directives and worked under the supervision of a registered physician who was responsible for all patient care on predetermined busiest periods for each ED.	Baseline 2 weeks	All ED patients: baseline n=9585; 2-week period 6 months postimplementation June 2007 n=10 007, of which PAs were on duty for 1076 visits and directly involved in n=376	Descriptive retrospective	Leaving without being seen Wait time (triage to initial assessment) Length of stay in ED	Ducharme (2009)33
Emergency medicine	To understand trends in emergency medicine and interprofessional roles in delivering this care […] The focus was on how doctors, PAs and nurse practitioners share emergency medicine visits.	USA National sample EDs of non-institutional general and short-stay hospitals in the 50 states and the District of Columbia from the National Hospital Ambulatory Medical Care Survey	PAs as providers of ED care and prescribers of medication in emergency medicine (7.9% of patients seen by PAs in 2004).	Physicians and nurse practitioners	Random sample of patient visits to hospital EDs (n=1 034 758 313), 1995–2004	Longitudinal	Proportion of visits in which medications are prescribed Mean number of prescriptions written per visit Non-narcotic analgesic prescriptions Narcotic analgesic/NSAID prescription by type of provider Patient contact growth by provider	Hooker (2008)34
Emergency medicine	To compare the analgesic practices of emergency physicians with that of PAs	USA ED within a suburban teaching hospital in Michigan with 90 000 annual visits	PAs were deployed for seeing patients presenting at the ED with isolated lower extremity trauma. PAs work closely with emergency physicians in the Prompt Care Area of the ED.	Emergency physicians	n=384 survey respondents of patients of all ages who presented at the ED with an isolated lower extremity injury evaluated with a foot or ankle radiograph, n=227 PA patients, n=153 emergency physician patients in a 9-week period	Prospective cohort	Analgesia prescribing	Kozlowski (2002)35
Emergency medicine	To evaluate PAs’ management of paediatric patients in a general ED through examination of the 72 hours’ recidivism rates of their younger paediatric patients	USA General urban ED treating approximately 58 000 patients annually, 20% of which are under 18 years	PAs evaluate, treat and discharge patients of any age independent of emergency physicians and PAs treating patients with consult from the emergency physician.	Attending emergency physician only	n=2798 PA only cases; n=984 PA with emergency physician; n=6587 emergency physician only	Comparative retrospective	72-hour revisits to the ED	Pavlik (2017)36
Emergency medicine	To compare the quality of ED pain management before and after implementation of the Joint Commission on the Accreditation of Healthcare Organizations’ standards in 2001	USA National sample EDs included in the National Hospital Ambulatory Medical Care Survey	The use of PAs in the care of patients presenting to the ED with a long bone fracture.	Patients presenting to the ED with a long bone fracture not seen by PAs (medical residents, internists)	n=2064 Patients presenting at the ED with a long bone fracture (femur, humerus, tibia, fibula, radius, or ulna) in two time periods: 1998–2000, n=834 of which 3% were seen by a PA, 9% by resident/intern and 90% by staff physician; 2001–2003 8% PA, 10% resident/intern, 90% staff physician	Retrospective cohort	Proportion of patients with long bone fracture receiving analgesia	Ritsema (2007)37
Emergency medicine	To compare the wound care practices and infection rates of wounds managed in the ED by practitioners with varying levels of medical training	USA Department of Emergency Medicine within a teaching hospital in New York	All patients with lacerations were evaluated by an attending physician who determined whether wound could be managed by a junior practitioner (PAs, students, interns, and residents).	ED patients whose wounds were managed by other providers (students, interns and residents)	All patients with lacerations attending the ED n=1163, n=901 seen by a PA, n=262 by other providers October 1992 to November 1993	Prospective observational	Patient wound infection rate	Singer (1995)38
Trauma and orthopaedics	To define the clinical and financial impact of hospital-based PAs on orthopaedic trauma care at a level II community hospital	USA Orthopaedic trauma care at a level II community hospital	Hospital-employed PAs (n=2) were used to cover all orthopaedic trauma needs, under the supervision of one of 18 orthopaedic surgeons. Each PA performed 12-hour day shifts for 3 consecutive days, January to December 2007. PAs on call carried trauma pagers and reported to the emergency room as soon as possible.	Attending surgeon as the primary orthopaedic responder for emergency department consults	n=1104 n=310: PA n=687: no PA	Comparative retrospective	Triage time to time seen by orthopaedic service in emergency department (min) Triage time to time of surgery (min) Operating room complication rates (%) The use of deep vein thrombosis prophylaxis (%) Postoperative antibiotic administration (%) Postoperative complications (%) Triage time to out of emergency department (min) Operating room set-up time (min) Average operating room time (min) Time from wound closure to wheels out (operating room) (min) Hospital length of stay (min) Cost savings (emergency department) ($) Cost savings (operating room) ($)	Althausen (2013)39
Trauma and orthopaedics	To describe the effect of PAs working in an arthroplasty practice from the perspective of patients and healthcare providers To describe the costs, time savings for surgeons and effects on surgical throughput and waiting times	Canada High-volume academic arthroplasty programme employing PAs (The Concordia Joint Replacement Group)	Addition of PAs (n=3) to the operating room team. The PAs were added to the team, replacing surgical assists (usually general practitioners). The PAs took first call with their supervising physician, provided first-assist services in the operating room (OR), write postoperative tests/investigations, generate operative notes, undertake daily working rounds and complete discharge summaries.	• Costs: GP first assists in the operating room • Waiting times: patients on the arthroplasty waiting list in 2004 and 2005	Sample size varying by outcome: • Patient satisfaction n=1070 • Perceptions of healthcare providers and patients n=44 • Costs n=402 surgical procedures performed in 2006 • Time savings n=1409 procedures carried out 2006 • Waiting times in 2006	Mixed methods	Patient satisfaction Perceptions of PAs among healthcare providers and patients Costs Time savings Waiting times Throughput	Bohm (2010)40
Trauma and orthopaedics	To assess whether the type of provider (attending physician vs PA) or number of providers involved in the non-operative management of a paediatric forearm fracture influenced the risk of that fracture healing as a malunion	USA Children’s hospital medical centre	PAs carrying out non-operative management of forearm fractures at orthopaedic clinic visits.	Attending physician	Patient charts of those aged 3–17 years seen at the orthopaedics department February 2012 to January 2013 n=141	Comparative retrospective	Fracture malunion (maximum angulation criteria) at last clinic visit	Garrison (2017)41
Trauma and orthopaedics	To describe the role of the PA in the upper extremity surgical programme; describe the role of the PA in an operating room study; and show the impact of the PA role on patients, providers and the system	Canada Subspecialised upper extremity surgical programme at a peripheral hospital, as part of a Physician Assistant Demonstration project where 12 PAs were introduced to various healthcare settings	One PA filling provider gaps in four areas: preoperative patient screening, assisting in operating room care (including a double-room experiment), aiding in aftercare of surgery and attending to postdischarge follow-up care.	Preoperative—surgeon working alone; operating room—team with surgical assistant or role unfilled and single operating room; surgery aftercare—replacing a postunfilled surgical extender; postdischarge—surgeon only	n=38 interviews; n=75 surveys (n=28 from healthcare providers and 47 from patients)	Mixed methods	Perceptions and experiences with the PA Patient rating of quality of care Expected and actual operating room times Total new patients seen	Hepp (2017)42
Trauma and orthopaedics	To assess whether staffing changes within a level 1 trauma centre improved mortality and shortened hospital and ICU length of stay for patients with trauma	USA Urban, community-based level I trauma centre	Core trauma panel (consisting of full-time, in-house trauma surgeons) and PAs	Group 1: general surgery residents (staffed by full-time, in-house postgraduate year 4 general surgery residents with attending back-up from home, followed by a transition to a trauma service staffed with in-house independent general surgeon attendings); group 2: core trauma panel (consisting of full-time, in-house trauma surgeons, without PAs or residents)	n=15 297 Trauma patients 18 years or older and not transferred from the ED to another acute care facility	Prospective cohort	Overall mortality Mortality for patients with injury severity score (ISS)>15 Hospital LOS	Mains (2009)43
Trauma and orthopaedics	To analyse patient outcomes and efficiency of care provided for trauma patients during transition from resident physician support to PA support	USA Level I trauma centre	PAs substituting for doctors in trauma alerts: PA’s role was to assist the trauma surgeon at trauma alerts and trauma patient rounds, update the trauma patient census list.	General and orthopaedic residents who attend in trauma alerts	n=293 before n=476-after All patients evaluated by the trauma surgeons and on the trauma registry, excluding those transferred to another facility for treatment of severe burns	Before-after	Collaborative relationship Transfer time LOS Mortality rate	Oswanski (2004)44
Internal medicine	To compare outcomes directly from the expanded use of PAs with those of a hospitalist group staffed with a greater proportion of attending physicians at the same hospital during the same time	USA Community hospital with 26 000 adult patients discharged annually	Expanded PA group: used three physicians and three PAs daily for ward rounds with PAs expected to see 14 patients daily plus one more PA responsible for day shift admissions. PAs worked in dyads with ward round physician; PAs discussed the treatment plans at least once a day with the physician to a written protocol for PA-physician dyad expectations.	Conventional group: used nine physicians and two PAs for rounding, with PAs expected to see nine patients daily, plus day shift admissions by the physician. PAs worked in dyads with ward round physician; PAs discussed the treatment plans at least once a day with the physician. No written protocol for PA-physician dyad expectations.	Patients discharged between January 2012 and June 2013; n=6612 expanded PA group and n=10 352 in the conventional group	Retrospective comparative	30-day all-cause readmission Inpatient mortality Cost of care Consultant/attending use Length of stay	Capstack (2016)45
Internal medicine	To examine and compare costs, between a PA service and an intern/resident (teaching) service in the provision of inpatient care for five high-volume internal medicine diagnostic-related groups	USA Two general internal medicine units, teaching hospital	The use of PAs (n=16) in the provision of care within internal medicine department (64 attending physicians on rotation coverage, scheduled to admit to either a PA or teaching service, with group assignment determined 1 year in advance).	The teaching service (32 interns/residents with an average experience of 1-year postmedical school)	Adult patients discharged in the following diagnostic-related groups: cerebrovascular accident/stroke, pneumonia, acute myocardial infarction discharged alive, congestive heart failure, gastrointestinal haemorrhage: n=923, of which n=409 PA and n=514 teaching service	Prospective cohort study	Relative value units (costs) Length of stay	Van Rhee (2002)46
Mental health	To examine the role of PAs in the care of patients with severe and persistent mental illness	Canada Assertive community treatment team, providing multidisciplinary care to patients with severe and persistent mental illness	A PA was hired to assist with intake psychiatric assessments, physical examinations, preventive care, and follow-up of psychiatric and medical complaints in a model of PA supervised by a psychiatrist.	No comparison	Assertive community treatment team members (three social workers, one psychiatrist, two psychiatric nurses, one occupational therapist, one recreational therapist, the PA)	Qualitative interview	Perceived effect and challenges of delivering psychiatric care with the PA model	McCutchen (2017)47

ED, emergency department; ICU, intensive care unit; LOS, length of stay; NSAID, non-steroidal anti-inflammatory drug; PA, physician assistant/associate.