Assessing the impact of anaesthetic and surgical task-shifting globally: a systematic literature review

Abstract The global shortage of skilled anaesthesiologists, surgeons and obstetricians is a leading cause of high unmet surgical need. Although anaesthetic and surgical task-shifting are widely practised to mitigate this barrier, little is known about their safety and efficacy. This systematic review seeks to highlight the existing evidence on the clinical outcomes of patients operated on by non-physicians or non-specialist physicians globally. Relevant articles were identified by searching four databases (MEDLINE, EMBASE, CINAHL and Global Health) in all languages between January 2008 and February 2022. Retrieved documents were screened against pre-specified inclusion and exclusion criteria, and their qualities were appraised critically. Data were extracted by two independent reviewers and findings were synthesized narratively. In total, 40 studies have been included. Thirty-five focus on task-shifting for surgical and obstetric procedures, whereas four studies address anaesthetic task-shifting; one study covers both interventions. The majority are located in sub-Saharan Africa and the USA. Seventy-five per cent present perioperative mortality outcomes and 85% analyse morbidity measures. Evidence from low- and middle-income countries, which primarily concentrates on caesarean sections, hernia repairs and surgical male circumcisions, points to the overall safety of non-surgeons. On the other hand, the literature on surgical task-shifting in high-income countries (HICs) is limited to nine studies analysing tube thoracostomies, neurosurgical procedures, caesarean sections, male circumcisions and basal cell carcinoma excisions. Finally, only five studies pertaining to anaesthetic task-shifting across all country settings answer the research question with conflicting results, making it difficult to draw conclusions on the quality of non-physician anaesthetic care. Overall, it appears that non-specialists can safely perform high-volume, low-complexity operations. Further research is needed to understand the implications of surgical task-shifting in HICs and to better assess the performance of non-specialist anaesthesia providers. Future studies must adopt randomized study designs and include long-term outcome measures to generate high-quality evidence.


Introduction
Evidence shows that there is a significant unmet 'surgical need' globally, with up to 5 billion people facing barriers to accessing surgical care and 2 billion people without any access (Funk et al., 2010).It is estimated that between 11% and 32% of the global burden of disease is surgical in nature (Debas et al., 2006), and >6% of all global avertable deaths and disabilities are preventable through the provision of essential surgery (Mock et al., 2015).These figures are likely to worsen in the future due to the epidemiological shift towards noncommunicable diseases and traumatic injuries and because of an ongoing maternal health crisis in low-resource settings (Luboga et al., 2009;Hoyler et al., 2014).
The shortage of trained workforce in anaesthesia and surgery is one of the primary barriers to patients receiving surgical care (Meara et al., 2015).According to the Lancet Commission on Global Surgery, countries with <20 surgeons, anaesthesiologists and obstetricians per 100 000 population may experience a health workforce crisis although this target is rarely met in resource-poor settings (Hoyler et al., 2014).Across 42 low-and middle-income countries (LMICs), national general surgeon density ranged between 0.13 and 1.57 per 100 000 population; obstetrician density was found to be between 0.042 and 12.5 per 100 000 population and anaesthesiologist density ranged between 0 and 4.9 per 100 000 population (Hoyler et al., 2014).This challenge also increasingly affects high-income countries (HICs), where rural catchment areas are often too small to support specialist physicians (Kornelsen et al., 2012;Grzybowski et al., 2013;Homan et al., 2013).Due to the paucity of education and training opportunities, combined with low pay and poor working conditions, anaesthesiologists and surgeons often opt to migrate internally, towards urban centres, or abroad.High rates of attrition lead to an inequitable distribution of the healthcare workforce, with rural and deprived areas experiencing significant gaps in surgical coverage (Hoyler et al., 2014;Henry et al., 2015;Falk et al., 2020).
To address this issue, the international community has widely advocated for the need to make more effective use of the existing healthcare workforce through task-shifting (WHO 2008;Luboga et al., 2009;Meara et al., 2015).This refers to the delivery of surgical procedures by individuals with shorter training and fewer qualifications than the specialist physicians who would normally perform surgeries.Surgical tasks can be delegated to two categories of healthcare professionals: non-physician clinicians (NPCs) and generalist physicians (non-specialists) (Federspiel et al., 2018;Falk et al., 2020).Task-shifting has been implemented across countries of all income groups: a review and survey conducted by Federspiel et al. (2015) identified 30 countries employing surgical task-shifting and 108 countries relying on anaesthetic task-shifting.
Despite the widespread adoption of this strategy, the evidence around its efficacy and acceptability remains fragmented and controversial.Its supporters primarily contend that task-shifting can ensure timely access to surgical care by expanding the pool of surgical providers and the skill-mix of operating teams (Dawson et al., 2014;Binda et al., 2021;Beard et al., 2022).For example, non-specialists perform 52.8% of all surgeries in Sierra Leone and 58.3% of procedures in Liberia (Bolkan et al., 2016;Adde et al., 2022).Other perceived advantages include its cost-effectiveness, given the lower training and remuneration costs (Kruk et al., 2007;Hounton et al., 2009;Beard et al., 2022), as well as higher retention rates of non-specialist clinicians (Chu et al., 2009).However, specialist physicians, patients and policymakers have questioned the quality of surgical care provided through task-shifting.They fear that this strategy could result in the creation of a two-tier healthcare system where patients who are operated on by specialists receive superior care compared with those treated by less specialized cadres (Ashengo et al., 2017;van Heemskerken et al., 2020).
The safety of anaesthetic and surgical care provided by non-specialists has been insufficiently evaluated, and it remains poorly understood (Federspiel et al., 2018;Schroeder et al., 2020).Our systematic review synthesizes the existing literature reporting on anaesthetic and surgical outcomes of patients who are operated on by non-physicians and nonspecialist physicians globally.As insufficient capacity is one of the main barriers to delivering surgical care worldwide, it is important to investigate the feasibility and safety of task-shifting to mitigate the shortage of healthcare providers.

Methods
We conducted a systematic literature review, in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al., 2009).A study protocol was written and registered on PROSPERO prior to commencing the review process.On 22 February 2022, we searched four electronic databases for relevant literature: MEDLINE, EMBASE, CINAHL and Global Health.To identify other relevant publications, we manually searched the bibliographies of all included articles and the reference lists of existing reviews on surgical task-shifting.
We developed a comprehensive search strategy, in conjunction with a university librarian with expertise in Health Policy.This included combinations of keywords and subject headings relating to surgical disciplines (i.e.surgery, anaesthesia and obstetrics), task-shifting and synonyms (i.e.task-sharing and task-delegation) and non-specialist healthcare professionals (i.e.non-physician, non-specialist and non-surgeon).Although we generally selected broad search terms pertaining to anaesthesia and surgery, without focusing on a particular surgical discipline, we chose to include keywords for caesarean section, laparotomy and open fracture management.This decision was made as these Bellwether procedures are indicative of functioning and comprehensive district-level surgical care delivery systems (Meara et al., 2015).The complete search strategy applied to MEDLINE can be viewed in Supplementary File 1.

Inclusion and exclusion criteria
Our systematic review focused on studies published from January 2008 until February 2022.This search period was chosen based on the launch of the first global conference on taskshifting, organized by the World Health Organization (WHO) in 2008.The searches were not restricted by language or by country setting.

Population
Study participants were patients who received anaesthesia or sedation as part of a surgical procedure or who underwent surgery.A surgical procedure was defined as 'the incision, excision or manipulation of tissue that requires regional or general anaesthesia or profound sedation to control pain' (Meara et al., 2015).Relevant procedures included both minor and major surgical operations, performed on urgent (emergency) or planned (elective) basis.Dental procedures and other medical services beyond anaesthesia and surgery were excluded.Patients of all ages and from all countries and healthcare settings were eligible for inclusion.

Intervention
Task-shifting was defined as the rational redistribution of anaesthetic and surgical tasks from highly qualified health workers to health workers with shorter training and fewer qualifications to maximize the available human resources for health (WHO, PEPFAR, UNAIDS, 2008).We focused on vertical task-shifting: anaesthesia and surgical tasks could be delegated from specialist surgeons or anaesthesiologists to non-specialist physicians or from physicians to NPCs.We included all studies in which non-specialists were the lead providers of anaesthetic or surgical care in the intervention arm.Given the feasibility in the scope of this work, it must be highlighted that we took a focused approach, concentrating on task-shifting in relation to the intraoperative performance of anaesthesia and surgery, rather than broader perioperative aspects of care (e.g.preoperative preparation or postoperative care) and non-technical skills (e.g.clinical decision-making).

Comparators
The comparator of interest was 'the standard practice': in the control group, anaesthesia and surgery were performed by the physicians who would ordinarily conduct the procedures in the study setting.Therefore, physicians could be specialists or non-specialists.A specialist was defined as a physician who had completed residency training in a surgical speciality or in anaesthesia (i.e. a qualified surgeon or anaesthesiologist).A non-specialist physician was as a medical doctor who had completed medical school but who did not have formal training in surgery or anaesthesia [i.e. general practitioner (GP)].Studies without a relevant comparator were excluded.

Outcomes
Studies that reported on anaesthetic or surgical outcome data were included.The primary outcomes of this systematic review were short-term and long-term mortality measures.Secondary outcomes comprised all other patient clinical outcomes because of anaesthetic and surgical task-shifting.We anticipated that common secondary outcome measures would include morbidity measures, length of hospital stay, unplanned hospital readmission, reoperation and patientreported outcomes.

Study designs
We included all randomized controlled trials (RCTs), quasiexperimental studies and prospective and retrospective observational studies and cross-sectional studies.Qualitative studies were excluded as the main clinical outcomes of interest centred around clinical safety, as opposed to patient perspectives and experiences of the anaesthetic and surgical care received.Existing literature reviews, case studies, conference abstracts and commentaries were not eligible.

Study selection
After removing duplicates in EndNote, two authors independently screened all results against the pre-specified inclusion criteria to identify eligible studies: first, they reviewed 'titles and abstracts' and, subsequently, the full texts of the selected articles were reviewed.Discrepancies were resolved through discussions and by involving a third reviewer until a consensus was reached.The screening process was managed through Rayyan.

Data extraction
The following data items were recorded for each study to establish combinability: the country setting and the healthcare facilities where patients were treated; the occurrence of either surgical or anaesthetic task-shifting and the characteristics of the procedures performed and the characteristics of surgical or anaesthetic providers, including the training received and the level of supervision of non-specialists.We also noted study designs and relevant clinical outcome variables.
In the event of missing data, we contacted study authors for clarification.All data points were independently extracted by one author into a pre-tested Microsoft Excel form; subsequently, a second author checked and confirmed the accuracy and completeness of the recorded information.

Risk of bias
The risk of bias of randomized studies was appraised through Version 2 of the 'Cochrane risk-of-bias tool for randomized trials' (RoB 2) (Sterne et al., 2019) (Tables 1 and 2).Domains included bias arising from the randomization process, bias due to deviations from the intended interventions, bias due to missing outcome data, bias in the measurement of outcomes and bias in the selection of reported results.Instead, we evaluated the risk of bias in non-randomized studies using the 'Risk

Data analysis
Given the heterogeneous nature of the interventions and outcomes assessed, we could not pool study data in a metaanalysis.Instead, we performed a narrative synthesis of results.To describe the findings, studies were grouped by taskshifting category (surgery or anaesthesia), location (LMICs or HICs) and surgical procedure type.

Search and screening results
Our initial search strategy identified 15 002 publications.Following duplicate removal, 9543 articles were screened at 'title and abstract' stage, and then, 131 full-text papers were reviewed.Two additional articles were identified through references, leading to a total of 40 studies being included in this systematic review.Further details are outlined in the PRISMA study flow diagram (Figure 1).

Characteristics of the included studies
Thirty-five studies exclusively addressed surgical task-shifting (Table 3) and four focused on anaesthetic task-shifting (Table 4).One study analysed both interventions (Cometto et al., 2012).Across all studies, the three most performed surgical procedures were caesarean sections (nine studies, 22.5%), male circumcisions (six studies, 15%) and hernia repairs (five studies, 12.5%).Thirty publications focused on delegation to NPCs, whereas in seven studies surgical or anaesthetic tasks were shifted to non-specialist physicians.In Chu et al. (2011) and Bolkan et al. (2017), the intervention group comprised both non-physicians and general medical doctors; Hounton et al. (2009) separately assessed the impact of task-shifting to both cadres.All studies were published between 2008 and 2021, with 24 papers (60%) published since 2015.Based on the World Thirty studies (75%) reported on perioperative mortality and 34 studies (85%) analysed morbidity measures, including perioperative complications, hospital length of stay and duration of the operative procedure.Fourteen articles (35%) adopted prospective study designs, including one RCT (Ashley et al., 2021), whereas 26 studies (65%) were retrospective.Twenty-two publications (55%) were multi-centre studies.

Obstetric surgery
In total, eight studies in sub-Saharan Africa assessed the impact of task-shifting for obstetric surgery: all papers compared outcomes of caesarean sections between non-physicians and physicians.In a prospective study by Gajewski et al. 2019b, wound infection rates for caesarean deliveries performed across eight district hospitals in Zambia were equivalent between medical licentiates and medical doctors.Gessessew et al. (2011) performed a retrospective study comparing caesarean delivery outcomes of 11 NPCs and 4 obstetricians in 13 Comprehensive Emergency Obstetric Care (CEmOC) facilities in Tigray Region, Ethiopia.NPCs performed 63.3% of all obstetric surgical interventions and 55.9% of emergency caesarean sections; they were associated with equivalent rates of maternal deaths, foetal deaths and hospital length of stay as specialist physicians.In Tanzania,  Ouédraogo et al. (2015) noted that surgical nurses were not inferior to obstetricians when comparing maternal mortality, maternal morbidity and neonatal mortality (2.6% vs 6.6%, P = 0.043).Similarly, Tariku et al. (2019) reported no significant differences in 'perinatal mortality and surgical complications' between non-physicians and physicians performing caesarean sections in two CEmOC hospitals in Ethiopia.In a study by (Van Duinen et al. 2019), despite higher readmission rates (3.7% vs 1.7%; OR 2.17, 1.08-4.42),caesarean sections performed by associate clinicians were shown to be safe, comparing favourably to physicians for maternal mortality (0.2% vs 1.8%; OR 0.11, 0.01-0.63)and stillbirths (12.7% vs 16.4%; OR: 0.74, 0.56-0.98).Blood loss, wound infection, postoperative pain and reoperation were not related to provider cadre and associate clinicians reported shorter operative times (31.9 vs 38.9 minutes, P < 0.001).McCord et al. (2009) reported no significant differences in mother or child fatal outcomes following emergency caesarean sections performed by assistant medical officers and medical officers in 14 hospitals across two regions of Tanzania.Additionally, there were no significant differences in major complications and quality indicators between the two patient groups.Instead, significant differences in patient outcomes were measured between government hospitals and mission hospitals.
In a retrospective, cross-sectional study in Burkina Faso, Hounton et al. (2009) compared the quality of caesarean sections performed by clinical officers, GPs and obstetricians.Although no significant differences were noted in maternal case fatality rates and maternal morbidity measures, newborn case fatality rates were higher for clinical officers as opposed to non-specialist and specialist physicians (198/1000 vs 125/1000 vs 99/1000).Furthermore, operative times and postoperative hospital length of stay were, respectively, 20% and 30% shorter for obstetricians.

Hernia repairs
In total, nine studies in sub-Saharan African countries addressed task-shifting for general surgery procedures.Most studies focused on elective or emergent hernia repairs: four studies compared the performance of non-physicians with medical doctors, whereas two studies compared non-specialist physicians and qualified surgeons.Ashley et al. (2021), the only RCT identified in this systematic review, compared the outcomes of hernia repair with mesh when performed by six associate clinicians and five medical doctors in Sierra Leone.At 1-year follow-up, there were no differences in mortality rates; solely one (0.9%) hernia recurrence was reported in the intervention group against seven (6.9%) in the control arm, demonstrating the non-inferiority of treatment by associate clinicians (P < 0.001).Similarly, there was an equivalent distribution of secondary outcomes at 2 weeks (postoperative complications) and 1 year post-surgery (fewer groin symptoms, pain, patient satisfaction and self-assessed health status).In a retrospective study across seven healthcare facilities in the Pwani Region of Tanzania, Beard et al. (2014) noted that clinical officers and assistant medical officers were the primary surgical providers in 87.1% of district hospitals and 67.2% of mission hospitals: overall, they performed 81.8% of elective hernia repairs and 81.2% of emergent hernia repairs.When comparing mortality and morbidity outcomes for NPCs and physicians, no significant differences between the two provider cadres were detected.Wilhelm et al. (2011) (retrospective) and Gajewski et al. (2019a) (prospective) studied the outcomes of inguinal hernia repairs undertaken by clinical officers and physicians in Malawi.Both papers concluded that there were no statistically significant differences in wound infection outcomes.Additionally, Wilhelm et al. (2011) reported comparable outcomes for mortality, anastomotic leakage, and reoperation rates and duration of hospital stay.
According to Beard et al. (2019), in a regional hospital in Ghana, three medical officers were deemed to be as safe as two general surgeons performing elective inguinal hernia repair with mesh.The former providers were noninferior based on mortality and hernia recurrence after 1 year (0.9% vs 2.8%, absolute difference −1.9, P < 0.01) as well as secondary outcomes at 2 weeks and 1 year post-surgery.Similarly, in a prospective study in Kenya, Wamalwa et al. (2015) found that medical doctors were equivalent to surgeons when undertaking hernia repairs using the Lichtenstein and the Shouldice techniques.Outcomes studied comprised postoperative complications at 24 hours and 2 weeks postsurgery, operative time and number of technically difficult operations performed.

Other general surgery
Beyond hernia repairs, Beard et al. (2014) found no significant differences in mortality and morbidity outcomes for other general surgery operations performed by NPCs and physicians.The most common procedures included prostatectomy, exploratory laparotomy and hydrocelectomy.In Gajewski et al. 2019b, wound infection rates for index general surgery procedures were equivalent for medical licentiates and medical doctors operating in Zambia.In addition to hernia repairs, Wilhelm et al. (2011) studied outcomes of ventriculo-peritoneal shunting (VP-shunting) for hydrocephalus and transvesical prostatectomy for benign prostate hyperplasia, which were led by clinical officers or by general surgeons in a referral and teaching hospital in Malawi.For VP shunting, rates of mortality and wound infection, early shunt revision and shunt explants were not statistically different between the two provider groups; however, average postoperative hospital stay was significantly shorter in the surgeon group (10 vs 8 days, P = 0.03).Analogously, postoperative hospital stay was longer for the intervention group (16 vs 15 days, P = 0.05) in the case of prostatectomy; no other outcome parameter differed between the two groups of operators (mortality, wound infection, bladder leakage or reoperation).Tobome et al. (2021) conducted a retrospective study in a peripheral, referral hospital in Western Benin where a GP and two general surgeons treated patients with acute generalized peritonitis.The GP performed 73% of operations, with mortality rates and postoperative complication rates analogous to the specialist physicians.In a retrospective study, Mpirimbanyi et al. (2017) assessed the clinical outcomes of all patients presenting to three rural district hospitals in Rwanda with emergency general surgical conditions.No significant differences between patients treated by a GP as opposed to a general surgeon were highlighted when assessing mortality, postoperative complications and hospital length of stay.

Surgical male circumcision
Four studies assessed outcomes of surgical male circumcisions undertaken by NPCs or physicians in sub-Saharan African settings (Buwembo et al., 2011;Rech et al., 2014;Ngcobo et al., 2018;Matumaini et al., 2021).All studies demonstrated the comparability of the lower-level cadres with the standard practice: there were no significant differences in the rates of perioperative adverse events (including bleeding, infection, wound dehiscence, swelling and pain) and in operative timings between provider types.Interestingly, Buwembo et al. (2011) highlighted that in a multivariate regression analysis surgical outcomes were significantly associated with provider experience, independently of professional title: increased surgical experience of the provider reduced operative duration by 1.5 minutes (P < 0.001).Attebery et al. (2010) reported on an assistant medical officer who was trained to provide essential neurosurgical services in a rural referral hospital in Tanzania.When comparing his performance with that of an American neurosurgeon, no differences in mortality rates were detected.Through a prospective observational study, Bolkan et al. (2017) sought to analyse the performance of trainees and graduates [Surgical Assistant Community Health Officers (SACHOs)] of a surgical training programme, promoted by the non-profit organization CapaCare and by the Ministry of Health in Sierra Leone.Adjusted mortality rates between 'indirectly supervised' and 'observed' surgical and obstetric procedures performed by trainees (0.8% vs 2.6%, OR 0.47, 0.32-0.71,P < 0.001) and by SACHOs (0.8% vs 9.6%, OR 0.16, 0.07-0.41;P < 0.001) proved the safety of task-shifting through this programme.Chu et al. (2011) described the outcomes of all surgical interventions taking place in a private hospital in Guri-El, central Somalia.Procedures were performed by specialist surgeons from Médecins Sans Frontières until January 2008 and by a local general doctor with surgical skills in conjunction with a surgical nurse between January 2008 and December 2009.Perioperative mortality was higher between 2006 and 2009 than in 2008-2009 (1.7% vs 0.2%, P < 0.001).In a study by Cometto et al. (2012), no statistically significant differences were found in mortality rates following 1543 surgical operations performed by surgical technicians or by visiting consultant surgeons in Primary Health Care Centres in Southern Sudan.Tyson et al. (2014) conducted a case-control study of paediatric surgery operations performed in a tertiary referral hospital in central Malawi by clinical officers as opposed to consultant surgeons and surgical residents.The NPCs led ∼40% of all cases, and they most commonly performed burn surgery, neurosurgery and ear, nose, throat procedures.Overall, mortality and complication rates stratified by case complexity were similar between clinical officers and specialists.The authors initially noted longer duration of hospital length of stay (24 vs 10 days, P < 0.001) and higher reoperation rates (17% vs 7.1%, P < 0.001) in patients operated on by clinical officers although these differences ceased after controlling for burn victims.Wilhelm et al. (2017) compared orthopaedic clinical officers and surgeons, performing major amputations and open reduction and internal fixations in Zomba Central Hospital, Malawi.The baseline characteristics of patients were comparable between provider cadres and no significant differences in mortality or postoperative complications emerged, including infection rates, postoperative blood transfusion, reoperation and postoperative hospital length of stay.

Other surgical procedures in LMICs
In an 8-year retrospective study in Melaka Hospital (Malaysia), Thevi et al. (2017) studied the relationship between the grade of operators and the occurrence of complications during cataract surgery: 11.7% of patients operated on by medical officers (non-specialists) experienced an intraoperative adverse event as opposed to 5.1% of patients treated by specialist ophthalmologists (P < 0.001).

Surgical task-shifting in HICs
Tube thoracostomy Two studies focused on tube thoracostomies in the USA.Bevis et al. (2008) sought to compare the quality of tube thoracostomies done by advanced practice providers (advanced registered nurse practitioners and physician assistants) and by trauma surgeons.Overall, no differences were identified in insertion complications, outcome complications requiring additional intervention or hospital length of stay.The only significant difference between practitioner types was 'complication when the thoracostomy tube extended caudad from the insertion site' (2.6% vs 21%, P = 0.02).No deaths occurred as a direct result of the tube thoracostomies.In a study by Harrell et al. (2020), thoracostomy tubes were placed by aeromedical non-physicians in a pre-hospital setting or by physicians in a hospital setting.There were no differences in mortality or complication rates.Although patients treated by non-physicians were more likely to be disposed to the intensive care unit (ICU) (67.3% vs 34.0%, P = 0.006) and experience significantly longer ICU length of stay (10.35 vs 6.70 days, P = 0.034), there were no differences in overall hospital length of stay, ventilator days and hospital disposition.Overall, in both papers, advanced practice providers were proven to be safe providers of tube thoracostomies.

Neurosurgery
Three publications discussed neurosurgical procedures.Enriquez-Marulanda et al. (2018) and Ellens et al. (2019) compared placements of external ventricular drains (EVDs) by trained advanced practice providers and physicians.Across both articles, the authors found no significant differences in placement accuracy, postoperative Glasgow Coma Score, and intraoperative and postoperative complications, including haemorrhage, infection and catheter leaks.In Ellens et al. (2019), 18 patients treated by non-physicians required multiple passes of the catheter for correct EVD placement, without further complications.Enriquez-Marulanda et al. (2018) observed that non-physicians had a higher risk of non-functional EVD (21.8% vs 11.9%, P = 0.04).Young and Bowling (2012) studied patient outcomes following intracranial pressure monitor positioning by mid-level providers or by neurosurgeons.The difference in major complication rates was significantly <5% (1.4% vs 0%, P = 0.0128); instead, there was a clinically significant difference in minor complication rates (5.7% vs 0%, P = 0.80).

Surgical male circumcision
Two studies from the USA focused on surgical circumcisions by non-physicians in the paediatric population.Gerber et al. (2019) did not identify any significant difference between advanced practice providers and urologists performing the procedure in terms of overall complications, revision of circumcision or need for penile surgery following circumcision and 30-day return to the emergency department .In Giramonti and Kogan (2018), operative times were comparable between advanced practice providers and urologists, with 63% of 'non-physician' cases and 46% of 'physician' cases taking 21-30 minutes to complete.Homan et al. (2013) sought to compare the outcomes of caesarean deliveries performed by family physicians or by obstetricians in two rural community hospitals in New England, USA.Although the family medicine hospital recorded longer average procedure times (55.2 vs. 42.5 minutes, P < 0.01) and maternal hospital length of stay (3.0 vs 2.6 days, P < 0.01), there was no evidence of increased risk for mothers and newborns.During the study period, no maternal deaths occurred in either healthcare facility and no statistically significant differences were found for intraoperative and infectious complications and newborn outcomes.Overall, mothers operated on by family physicians experienced fewer maternal postoperative complications (0.03 vs 0.12, P = 0.03).

Basal cell carcinoma excision
Finally, Ramdas et al. (2018) analysed pathology records of excisions of basal cell carcinomas performed by GPs as opposed to dermatologists or plastic surgeons in the Netherlands between 2008 and 2014.The rates of complete excisions were significantly lower for GPs as opposed to the dermatologists (70% vs 93%, P < 0.001), with the risk of incomplete excision being six times higher for procedures completed by a GP (OR 6.2,, P < 0.0001).

Anaesthetic task-shifting globally
Five papers implemented a comparative approach to study patients' clinical outcomes following the administration of anaesthesia or sedation for surgical procedures.Three studies were set in LMICs and two in the USA.All studies compared the performance of non-physicians with that of qualified anaesthesiologists.Cometto et al. (2012) reported on anaesthesia provided by anaesthesia technicians (non-physicians) and by consultant anaesthetists during surgical missions of the NGO Comitato Collaborazione Medica in Southern Sudan.Operations were mostly undertaken under spinal anaesthesia (60%), whereas ketamine and general anaesthesia with endotracheal intubation were used in a minority of cases.The total mortality rate was 0.58%, and no significant differences between provider groups were identified.Kudsk-Iversen et al. (2020) examined the safety of anaesthesia care provided by Médecins Sans Frontières in crisis settings across 23 countries between January 2008 and December 2017.In active armed conflict situations, mortality rates were 0.3%, 0.2% and 0.3% for uncertified providers, nurse anaesthetists and anaesthesiologists, respectively; in poor-resourced contexts experiencing healthcare gaps, the mortality rates were 0.2% for uncertified providers, 0.1% for nurse anaesthetists and 0.3% for physician anaesthesiologists (van der Merwe et al., 2021) sought to describe the outcomes of 336 procedural sedations for adult elective and emergency surgery, administered by non-physician or physician anaesthesia providers across 25 African countries.Through an inverse probability of treatment-weighted model, the authors estimated the average treatment effect of exposure to a non-physician provider: they detected an 8-fold increase in the odds of 'severe complications and death' in patients treated by non-physicians (12.8% vs 1.6%; adjusted OR 8.3, 2.7-25.6).
Two publications were set in the USA.Needleman and Minnick (2009) studied four outcome measures to establish the safety of nurse anaesthetists providing anaesthesia for obstetric surgery.Taking the 'anaesthesiologist-only' group as the reference category, nurse anaesthetists operating independently were not inferior based on mortality, anaesthesia complications, other complications and obstetrical trauma indicators.Analogous results were shown for nurse anaesthetists operating under the supervision of physicians.Dulisse and Cromwell (2010) assessed the clinical outcomes of patients who underwent anaesthesia between 1999 and 2005 in states where nurse anaesthetists could operate without supervision (opt-out states).The reference group was the 'anaesthesiologist solo group' in non-opt-out states.Nurse anaesthetists operating independently in opt-out states were not inferior to the control group for mortality (OR 0.689, P = 0.05) and complications (OR 0.813, P = 0.05).

Discussion
Our review identified 40 studies assessing patient outcomes because of anaesthetic and surgical task-shifting published between 2008 and 2021.All but four of the studies focused on surgical task-shifting, and the majority were conducted in sub-Saharan Africa.Only one study was an RCT, highlighting the limited high-quality evidence on the clinical implications of task-shifting.
In LMICs, the body of evidence mostly pertains to caesarean sections, general surgery, including hernia repairs, and surgical male circumcisions.Overall, evidence indicates that these common and low-complexity procedures have been performed safely by trained non-physicians and non-specialist physicians, with comparable outcomes between intervention and control groups.This aligns with existing literature that compared caesarean sections, hernia repairs and male circumcisions between non-physician and physician operators, without reporting significant differences in key outcomes (Wilson et al., 2011;Ford et al., 2012;Schroeder et al., 2020;Zakhari et al., 2022).However, we identified limited literature addressing surgical task-shifting in HICs and a lack of information on anaesthetic task-shifting with findings being generally of poor quality, rendering it precocious to draw firm conclusions.In LMICs, the current evidence on anaesthetic taskshifting appears to be conflicting, as previously highlighted by Lewis et al. (2014) and Sobhy et al. (2016).Even though two studies demonstrated similar outcomes between provider groups (Cometto et al., 2012;Kudsk-Iversen et al., 2020), results from a large prospective cohort study conducted across 25 African countries suggested that performance of nonanaesthesiologists was inferior to that of anaesthesiologists (van der Merwe et al., 2021).

Strengths and limitations
To the best of our knowledge, this is the most comprehensive literature review with a broad scope to synthesize the evidence of intraoperative anaesthetic and surgical task-shifting across all World Bank income groups.We employed a broad eligibility criterion with a comprehensive search strategy across four databases to capture all the existing literature examining the clinical impact of task-shifting to non-physicians or non-specialist physicians.
Our search strategy excluded any articles published prior to 2008, possibly omitting some research studies on the clinical implications of task-shifting.However, while task-shifting may have been studied in certain settings before, we focused on a period with sustained policy focus following the WHO's first conference on task-shifting (WHO, PEPFAR, UNAIDS, 2008).A methodological limitation of this study relates to its exclusion of broader perioperative aspects of surgical care provision, beyond the technical performance of anaesthesia and surgery.
Due to the overall poor quality of the included studies, and their heterogeneous nature, we were precluded from performing a meta-analysis of the extracted data.Nonetheless, our narrative synthesis offers broad insights into the clinical areas and the settings where task-shifting has been adopted, with important lessons for policy and practice.Since many studies employed non-randomized study designs, often with poor comparability between intervention and control arms, most articles in this review were deemed to be at moderate or serious risk of bias.Many studies failed to account for patient, procedure and healthcare setting characteristics, leading to imbalances between patient groups.For instance, specialists may have managed patients at higher risk of experiencing poorer outcomes and performed the most complex operations.It is also possible that non-specialists may have been more likely to be stationed in rural and public hospitals, which are often underresourced and lack accessibility to life-saving equipment or appropriate referral systems.
It was not uncommon for retrospective studies to report high rates of missing outcome data caused by the difficulty in retrieving patient files and their incompleteness.In studies where non-specialists operated under supervision, the direct involvement of physicians in the procedures often remained unclear, possibly leading to biased results.Finally, many studies included a short follow-up period, often measuring perioperative or in-hospital mortality and morbidity rates.However, anaesthetic and surgical complications are likely to manifest over time.Our review has also highlighted that there is a substantial gap in the literature responding to the research question beyond sub-Saharan Africa and the USA.This is not surprising, as surgical task-shifting is illegal in many parts of the world, but it is common practice in many countries across Africa.

Implications for future research
Further research is needed to understand the implications of surgical task-shifting in middle-income countries and HICs and to better assess the performance of non-specialist anaesthesia providers, given the scarcity of available literature.Studies on surgical task-shifting should adopt a broader lens, performing a robust analysis of all aspects of perioperative care, beyond non-specialists' technical performance in the operating theatre.It remains essential to evaluate the care provided by non-specialists across the perioperative pathway, including preoperative and postoperative management.Focus must also be placed on non-technical skills, such as clinical decision-making, communication and situational awareness.To strengthen the quality of the available evidence, future studies should adopt randomized study designs and include long-term outcome measures to generate high-quality evidence.

Policy implications
Our results suggest that task-shifting in low-complexity procedures such as caesarean sections and hernia repairs may be implemented safely.This is an important finding, given the high unmet need in obstetrics and general surgery in many low-resource settings (Bolkan et al., 2016;Lindheim-Minde et al., 2021).However, the effective adoption of task-shifting requires a modification of existing regulatory and governance systems, with formal recognition by Ministries of Health and professional bodies within medical practice regulations and national workforce strategies (Baine and Kasangaki, 2014;Ashengo et al., 2017;Jumbam et al., 2022).Evidence-based guidelines are required to define the precise scope of practice of non-physicians and non-specialist physicians and to establish credentialed pre-serving training curricula, ensuring that clinicians can operate safely and within a well-defined legal framework.This may include the incorporation of task-shifting as a workforce development strategy to address surgical need into National, Surgery, Obstetrics and Anaesthesia Plans in healthcare systems across low-, middle-and high-income country settings.Anaesthetic and surgical taskshifting policies must also be accompanied by systems of prospective recording and reporting, as well as the development of regulatory systems for quality assurance.It is essential to rigorously monitor long-term outcomes across all countries where task-shifting is implemented to evaluate the performance of lower-level cadres and to ensure accountability (Schroeder et al., 2020).
Training programmes should be recognized as an indispensable component of successful task-shifting initiatives.These enable the extended surgical team to develop technical skills as well as non-technical skills, including decisionmaking, effective communication and situational awareness.Importantly, training ought to be tailored to local contexts, preparing trainees for the challenging working conditions, which they would face in poorly resourced settings (Edgcombe et al., 2018;Kudsk-Iversen et al., 2018;Igaga et al., 2021).After qualification, supervision, interprofessional collaboration and continuing education are integral to enable anaesthesia and surgical providers to maintain and strengthen their competencies over time (Edgcombe et al., 2018;Kudsk-Iversen et al., 2018).To guarantee sustainability, one option could be to investigate how digital telecommunication technologies and e-learning platforms can be leveraged in resource-constrained settings to better connect specialists and non-specialist providers and facilitate access to professional development opportunities (Drum and Bould, 2017;Igaga et al., 2021;Mwapasa et al., 2021).Task-shifting cannot substitute specialist physicians, who remain integral to performing complex procedures, making informed clinical decisions and training surgical care providers (Beard et al., 2014;Kudsk-Iversen et al., 2018).In line with an interprofessional approach, task-shifting must be implemented alongside investments in 'standard' speciality training programmes for anaesthesiologists and surgeons.Both policy approaches are not mutually exclusive and should be developed contemporarily and complementarily to strengthen anaesthesia and surgical delivery globally.
Although task-shifting has been extensively advocated as a solution for the human resources crisis in global surgery, some studies have suggested that trained non-specialists may also be prone to leave the areas in greatest need to seek personal opportunities elsewhere (Schneeberger and Mathai, 2015;Gajewski et al., 2019a;Asefa et al., 2020).For example, Asefa et al. (2020) reported attrition rates as high as 65.9% and 71.4% for non-specialist surgical and anaesthetic personnel 15 months after deployment to rural hospitals.It remains crucial to research incentives to upkeep healthcare workers' motivation and facilitate their retention.Financial measures to attract non-specialist staff might include higher salaries coupled with bonding schemes, performance-based financing and hardship allowances together with social amenities, such as housing subsidies, health insurance and well-performing schools for their families.Furthermore, recognition, programmes for further training and specialization, clear career pathways and opportunities for professional growth appear to be indispensable to favour job satisfaction (Baine and Kasangaki, 2014;Nyawira et al., 2022).

Conclusion
Research predominately from sub-Saharan Africa suggests that non-specialists have performed common and lowcomplexity procedures, including caesarean sections and general surgery, with comparable outcomes to specialist physicians, thereby increasing access to safe and timely essential surgery.However, there remains a lack of high-quality research on the safety of surgical task-shifting for more complex procedures in LMICs and to further study the adoption of this strategy in HICs.More focus must be placed on anaesthesia care to better assess the performance of non-specialist providers in this field.In all contexts, for task-shifting to be sustainable and effective, policy efforts must centre around its formal recognition in regulatory and governance systems, around strengthening adequate training and supervision programmes and investigating strategies to favour the retention of healthcare workers in resource-poor contexts.

Figure 1 .
Figure 1.The PRISMA study flow diagram

Table 1 .
ROB-2, risk of bias judgements for randomized studies

Table 2 .
ROBINS-I, risk of bias judgements for non-randomized studies

Table 2 .
(Continued) (Sterne et al., 2016)ized studies of interventions' instrument (ROBINS-I)(Sterne et al., 2016).We assessed bias due to confounding, selection of participants into the study, classification of the interventions, deviations from the intended interventions, missing outcome data and measurement of outcomes.

Table 3 .
Characteristics of included studies, surgical task-shifting

Study name, country Study design Health- care facility characteristics Procedures Lead surgical providers and number of procedures Training and level of supervision of non-specialists Outcomes assessed Mortality Other clinical outcomes
(continued) Table 3. (Continued) Study name,

Table 4 .
Characteristics of included studies, anaesthetic task-shifting