Cochrane Review Says “Little Difference” Replacing Hospital Physicians with Nurses: We Disagree

Response to

“Substitution of nurses for physicians in the hospital setting for patient, process of care, and economic outcomes”
Bulter M, Kirwan M, McCarthy V, Cole J, Schultz T. (2024)

Physicians for Patient Protection
Research Committee Co-Chairs: Rebekah Bernard, MD; Phil Shaffer, MD

Physicians for Patient Protection welcomes this recent review examining substitution of nurses for physicians in hospital settings. We are encouraged to see growing scholarly interest in comparing outcomes between nurse practitioners and physicians, particularly as scope-of-practice policies continue to evolve in the United States.

However, after careful review of the 82 studies included, we believe the title and framing of this review overstate the evidence for true inpatient substitution.

1. “Hospital Setting” — But Only Seven Inpatient Studies

Although the review is titled “hospital setting,” only seven of the 82 studies evaluated inpatient care:

• Goldie (2012) – Canada
• Griffiths (2000) – UK
• Griffiths (2001) – UK
• Mitchell-DiCenso (1996) – Canada
• Rosenthal (2001) – USA
• Steiner (2001) – UK
• Walsh (1999) – UK

Even within these inpatient studies, none evaluated independent nurse-led hospital units. All occurred within physician-led systems and under physician supervision.

Thus, the review does not demonstrate nurses independently managing hospital floors or critically ill patients without physician oversight.

2. Limited and Outdated U.S. Evidence

Only six of the 82 studies were conducted in the United States:

• Rosenthal (2001)
• Pioro (2001)
• Two endoscopy studies (see below)
• Two studies examining added nursing services

The most recent U.S. study included was published in 2011, meaning there is no contemporary U.S. hospital-based substitution research reflecting current NP education pathways.

Given the expansion of fully online programs and direct-entry NP models since 2011, this absence of updated data is highly relevant.

3. Evaluation of the Six U.S. Studies

A. Studies Evaluating “Additional Nursing Services,” Not Substitution

Two studies examined whether adding nursing services improved outcomes — not whether nurses and physicians perform the same clinical role equivalently.

• In an emergency department study of chest pain patients (Allison 2000), registered nurses provided additional post-discharge counseling (lifestyle modification and risk factor management). Patients who received these added services showed improved risk factors compared to usual care. This finding is unsurprising; adding resources typically improves outcomes. However, the study did not compare nurses and physicians performing the same role
• In another study, resident physicians were either told where fecal occult blood testing supplies were located or instructed to refer patients to a nurse clinician; referral increased testing rates (Cargill 1991). This compares types of instruction—not the clinical performance of nurses versus physicians completing the same task. In our view, a true substitution study would require both clinician types performing the same clinical function under comparable conditions.

B. Pediatric Trauma Team Study

The pediatric trauma study (Fanta 2006) compared two attending physician–led teams:

• One included residents.
• One included nurse practitioners.

Limitations:

• NP team admitted Monday–Thursday.
• Resident team admitted Friday–Sunday.
• Resident team managed ~50% more patients.
• Resident patients had higher injury severity scores.

Weekend admissions are associated with worse outcomes and longer length of stay (well-established “weekend effect” literature).

Without controlling for admission timing and injury severity, the results cannot isolate team composition as the causal factor.

Importantly, both teams were physician-led.

C. Rosenthal et al. (2001)

Rosenthal (2001) compared two physician-led teams:

• One with residents
• One with nurse practitioners

Notable findings:

• NPs did not work nights; overnight care was physician-provided.
• 47% of patients assigned to the NP team were transferred to the physician-only team.
• Transfers occurred when patients were considered “too sick” or anticipated to require overnight physician monitoring.

This crossover suggests practical recognition of limits in NP-only coverage for more complex patients.

Again, this was not physician substitution.

D. Endoscopy Studies

Two U.S. procedural studies directly compared nurses or nurse practitioners with physicians:

• Limoges-Gonzalez (2011) – colonoscopy comparison
• Schoenfeld (1999) – nurse endoscopist trial

Key context:

• Nurses completed 100–140 supervised procedures prior to independent practice.
• In one study, the NP completed fellowship-style training similar to gastroenterology fellows.
• That NP had performed approximately 1,000 colonoscopies before study initiation.

These findings show that when nurses receive training comparable to physician procedural preparation for a defined task, they may perform that task competently.

However, such training is not representative of standard NP education.

Non-USA studies- Protocol-Driven Care Is Not Independent Substitution

The remainder of the studies in Cochrane were performed outside the US.  In addition to mostly studying non-hospital based services,

Many studies involved nurses following strict protocols and sometimes guided by decision-support software:

• Diabetes management (Arts, 2012)
• Pediatric gastroenterology constipation clinic (Burnett, 2004)
• Childhood asthma protocol (Kamps, 2003)
• Ventilator weaning (Fagoni, 2018; Rushforth, 2005)
• Blood pressure management post-stroke (Wahab, 2017)
• Fibromyalgia clinic (Kroese, 2008)
• ACE-inhibitor and Beta-blocker titration following a heart attack (Oyanguren, 2021)

In several cases, in addition to following protocols, Cochrane’s notes reflect that physicians supervised care:

• Atrial fibrillation clinic (Hendriks, 2012)
• Rheumatoid arthritis clinic (Kwok, 2022)
• Bronchiectasis clinic (Sharples, 2002)
• Early medical abortion under physician oversight (Warriner, 2011)

These represent protocolized task delegation within physician-directed systems — not independent physician replacement.

5. Telephone Follow-Up and Telemonitoring

Several studies involved structured telephone follow-up, in the same way that nurse triage is routinely performed:

• Beaver (2009; 2016) – cancer follow-up
• Bergsten (2019) – oncology
• de Thurah (2018) – rheumatoid arthritis
• Willems (2008) – asthma telemonitoring
• Kumar (2019) – hepatitis B telemonitoring

These involved stable patients with structured scripts and predefined escalation pathways.

They do not reflect inpatient substitution.

6. Minor or Straightforward Conditions

Studies of minor conditions included:

• ED fast track (Dinh, 2012; Jennings, 2015), studied how quickly pain medicine was given
• Soft tissue injury (McClellan, 2012)
• Minor injuries (Sakr, 1999)
• Eczema or atopic dermatitis (Moore, 2009; Schuttelaar, 2010)

These represent management of low-acuity or self-limited conditions.

7. Stable Chronic Conditions and Junior Doctor Comparisons

Some studies compared nurses to interns or junior doctors:

• Rheumatoid arthritis / osteoarthritis (Hill, 1994; 2003; 2009)
• Anticoagulant management (Vadher, 1997)
• Stable COPD (Vrijhoef, 2007)
• Antiretroviral therapy (Sekiziyivu, 2021)

Others involved providing education:

• Cardiogenetics (Nieuwhof, 2017)
• Epilepsy education (Ridsdale, 2000)

These do not represent substitution for fully trained attending physicians managing acutely unstable hospitalized patients.

Conclusion

After reviewing the evidence cited in Bulter et al. (2024), several points are clear:

1. Only 7 of 82 studies evaluated inpatient care.
2. Even those inpatient studies were physician-led or physician-supervised.
3. The majority of studies reviewed by Cochrane involved:
   • Protocol-driven care
   • Stable or low-acuity patients
   • Minor conditions
   • Telephone or outpatient follow-up
4. There is almost no contemporary U.S. hospital-based substitution research.

No study demonstrated nurses independently running hospital floors or ICUs without physician oversight.

As scope-of-practice policies expand, high-quality contemporary U.S.-based research examining true inpatient substitution is urgently needed. The title of the Cochrane report is misleading at best and may give the erroneous impression that nurses independently substitute for physicians in hospital settings—an assertion not supported by the evidence reviewed. Policymaking should be guided by rigorous evidence reflecting current educational standards and real-world hospital complexity.

We hope this review serves as a catalyst for more precise and methodologically rigorous study.