Substitution of nurses for physicians in the hospital setting for patient, process of care, and economic outcomes

RATIONALE: The demand for health services to deliver hospital-based care has increased due to an ageing population, more complex healthcare needs, comorbidities, and increasing healthcare costs. Nurse-physician substitution can improve access to care for patients who may otherwise have a significant wait for review by a physician. OBJECTIVES: The main objective of this review was to examine the impact of substituting nurses for physicians in the hospital setting (hospital inpatient units and outpatient clinics) on patient outcomes, process of care outcomes, and economic outcomes. The secondary objectives of this review were to assess whether the effects of nurse-physician substitution differ according to healthcare setting (low- and middle-income countries (which included low-income, lower middle-income, and upper middle-income countries) versus high-income countries), patient type, patient disease, intervention type (inpatient care, nurse-led clinics, role substitution, and task substitution), nurse grade, additional training, level of responsibility, and mode of substitution for nurse-led clinics (telephone/telehealth, partial substitution, enhanced substitution, and full substitution). SEARCH METHODS: We searched CENTRAL, MEDLINE, Embase, NHSEED, CINAHL, ProQuest, two citation indexes, and two trial registries. We also conducted handsearches, reference checking, and contacted study authors to identify eligible studies. We searched five grey literature databases and contacted experts relevant to the review area. The evidence is current to 25 June 2024. ELIGIBILITY CRITERIA: We included randomised controlled trials of both individual and cluster design that compared the effects of care delivered by a nurse to that delivered by a physician on patient, process of care, and economic outcomes. We included care of inpatients or outpatients in hospital settings. OUTCOMES: Our critical outcomes were mortality, patient safety events, and clinical outcomes. Important outcomes were quality of life, self-efficacy, relative performance of practitioner, and direct costs. RISK OF BIAS: We assessed the risk of bias using the Cochrane RoB 2 tool. SYNTHESIS METHODS: We synthesised results for six outcomes (70 studies) using meta-analysis. We calculated odds ratios (OR) or risk ratios (RR) with 95% confidence intervals (CI) for dichotomous outcomes, and standardised mean differences (SMD) or mean differences (MD) with 95% CI for continuous outcomes. Data for clinical outcomes and relative performance of practitioner were stratified, and direct costs were summarised narratively. We assessed the certainty of evidence using GRADE for all outcomes. INCLUDED STUDIES: We included 80 parallel-group single or multi-site randomised studies and two cross-over single-site randomised studies. There were 28,041 participants with study sample sizes ranging from 7 to 1907 participants. Studies included specialist nurses, advanced nurse practitioners, and registered nurses substituting for junior and senior doctors across a range of specialties, including: cancer, cardiology, dermatology, endocrinology, gastroenterology, general medical, neonatology, neurology, obstetrics/gynaecology, ophthalmology, rheumatology, and respiratory. Nurses practised autonomously or under the supervision of a doctor, sometimes using protocols and sometimes providing care for specific groups of patients. Most studies lasted for 12 months, with two studies lasting up to five years. The studies were carried out worldwide, with the majority being conducted in the UK (39%). Seventy-two studies were conducted in high-income countries. We judged 34% of the studies as having low risk of bias, and 46% as having some concerns. SYNTHESIS OF RESULTS: Meta-analyses showed there is probably little to no difference between nurse and physician care in mortality (RR 1.03, 95% CI 0.87 to 1.21; I² = 0%; 19 studies, 8239 participants; moderate-certainty evidence), quality of life (SMD 0.10, 95% CI -0.04 to 0.23; I² = 65%; 22 studies, 5246 participants; moderate-certainty evidence), and self-efficacy (SMD 0.01, 95% CI -0.06 to 0.09; I² = 0%; 11 studies, 3022 participants; moderate-certainty evidence), and that there may be little to no difference in patient safety events (RR 0.92, 95% CI 0.84 to 1.01; I² = 9%; 31 studies, 14,437 participants; low-certainty evidence). We did not pool results for clinical outcomes overall (36 studies, 5177 participants), but found there may be little to no difference between nurse-physician substitution and physician-led care for most clinical outcomes. However, for a small number of outcomes, nurse-physician substitution may result in improvements in clinical outcomes. We did not pool results for relative performance of practitioner overall (22 studies, 13,818 participants), and found that results for some outcomes may favour nurse-physician substitution; some may favour physician-led care; and for several outcomes there may be little to no difference between the two. The certainty of evidence was low for direct costs. A narrative synthesis of direct costs (36 studies, 15,230 participants) showed 17 studies reporting reduced costs when physicians were substituted with nurses, with nine studies reporting increased costs owing to longer consultations, referrals, and drug prescription. The risk of bias was generally low across studies. AUTHORS' CONCLUSIONS: In our review, we found little to no difference between nurse-physician substitution and physician-led care. Although nurse-physician substitution may result in better outcomes in certain cases, the evidence is uncertain. In considering nurse-physician substitution as a solution to physician shortages, we also need to consider its impact on the nursing workforce. FUNDING: No financial support was received. REGISTRATION: Protocol (2020) DOI:10.1002/14651858.CD013616.