Evidence, not eminence, for surgical management during COVID-19: a multifaceted systematic review and a model for rapid clinical change

Abstract Background Coronavirus (COVID-19) forced surgical evolution worldwide. The extent to which national evidence-based recommendations, produced by the current authors early in 2020, remain valid, is unclear. To inform global surgical management and a model for rapid clinical change, this study aimed to characterize surgical evolution following COVID-19 through a multifaceted systematic review. Methods Rapid reviews were conducted targeting intraoperative safety, personal protective equipment and triage, alongside a conventional systematic review identifying evidence-based guidance for surgical management. Targeted searches of PubMed and Embase from 31 December 2019 were repeated weekly until 7 August 2020, and systematic searches repeated monthly until 30 June 2020. Literature was stratified using Evans’ hierarchy of evidence. Narrative data were analysed for consistency with earlier recommendations. The systematic review rated quality using the AGREE II and AMSTAR tools, was registered with PROSPERO, CRD42020205845. Meta-analysis was not conducted. Results From 174 targeted searches and six systematic searches, 1256 studies were identified for the rapid reviews and 21 for the conventional systematic review. Of studies within the rapid reviews, 903 (71.9 per cent) had lower-quality design, with 402 (32.0 per cent) being opinion-based. Quality of studies in the systematic review ranged from low to moderate. Consistency with recommendations made previously by the present authors was observed despite 1017 relevant subsequent publications. Conclusion The evidence-based recommendations produced early in 2020 remained valid despite many subsequent publications. Weaker studies predominated and few guidelines were evidence-based. Extracted clinical solutions were globally implementable. An evidence-based model for rapid clinical change is provided that may benefit surgical management during this pandemic and future times of urgency.


Introduction
Coronavirus (COVID-19) has challenged surgical practice worldwide 1 , forcing staff to adapt and departments to restructure 2 . Surgery on patients infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) produces poor outcomes 3,4 making appropriate preoperative screening a necessity 5 . Transmission capability 6 and biodistribution 7 of the virus threaten the safety of operating room staff, especially those exposed to surgical smoke and aerosol-generating procedures 8,9 . Disruption associated with the pandemic may burden surgical systems for close to a year 10 , despite vaccination at population level and adaptations in management to cope with SARS-CoV-2 mutations.
Ideological differences in both operative and non-operative delivery of care existing within the global surgery community 11 have increased since COVID-19. This heterogeneity was reflected in the early operative recommendations, based mainly on expert opinions, and left considerable uncertainty 12,13 . Past consensus collaborations with a common philosophy have improved surgical safety worldwide 14,15 , and there is need for similar efforts to maintain effective surgical care during the COVID-19 era 16 .
Evidence-based principles have previously guided surgical innovation at an international level 17 . In April 2020, observing the need for rapid clinical change in response to COVID-19, the authors used this philosophy to produce guidance for the urgent adaptation of surgical services on a national scale 18 . Targeted rapid reviews were combined with the advice of clinical experts to produce evidence-based recommendations for three aspects of practice: safety of intraoperative practice for open versus laparoscopic surgery 19,20 , use of personal protective equipment (PPE) 21,22 and surgical triage 23,24 . As time has passed and the COVID-19 literature has grown 25 , it is unclear if these recommendations remain valid.
Although reliable models exist for clinical change in surgical management 17 , none achieve this in the rapid fashion required for urgent circumstances. As the COVID-19 pandemic has forced a uniquely rapid series of modifications to surgical services worldwide, understanding changes in the surgical literature since the identification of SARS-CoV-2 26 may benefit the development of a model for rapid evidence-based adaptations of surgical services.
To inform surgical practice during the COVID-19 era with an evidence-based model for rapid clinical change in surgical management, a multifaceted systematic review was conducted. This aimed to characterize the evolution of surgery since the identification of SARS-CoV-2, through rapid reviews targeting three aspects of surgical care (intraoperative safety 19,20 , PPE 21,22 and surgical triage 23,24 ) complementing a conventional systematic review of published evidence-based guidance for surgical practice during the pandemic.

Methods
The rapid data sharing that followed the COVID-19 outbreak 25 prompted a multifaceted approach to systematic review 1 . All searches were staggered temporally to identify subtle updates to the evidence base. Both rapid review and conventional systematic review methodologies were utilized in a complementary fashion within a multifaceted approach. The same rapid review methodology used during April 2020 [19][20][21][22][23][24]27 was implemented to characterize the evolution in the surgical literature during the pandemic. Searches targeting the three aforementioned areas of interest [19][20][21][22][23][24] were repeated at approximately weekly intervals. In contrast, conventional systematic review methodology was repeated at monthly intervals to evaluate the quality and evolution of published guidance for surgery derived from a formal literature search.
No language or publication restrictions were applied. The research questions and inclusion criteria were established a priori. As some articles were published in preprint or 'in press' form in addition to their final publication format 25 , searches were date-restricted by date of database entry rather than date of publication. Duplicates were removed when identified, with the earliest record retained. Searches were supplemented by consultation of current reviews and original research relating to surgery during the COVID-19 pandemic identified through targeted searches of PubMed and Google Scholar.

Rapid reviews
Searches sought to identify studies of any design, in any setting, directly relevant to three surgical topic areas: safety of intraoperative practice (including endoscopy procedures of the gastrointestinal or respiratory tract) 19,20 , use of PPE 21,22 and surgical triage 23,24 . Due to the similarity in PPE requirements for surgical and non-surgical staff, studies relevant to PPE in non-surgical settings were also sought. Six search strings were developed targeting safe intraoperative practice 19,20 , PPE 21,22 and surgical triage 23,24 in PubMed (incorporating MEDLINE) and Embase. These six searches were retrospectively conducted four times per month at approximately weekly intervals from 31 December 2019 (identification of SARS-CoV-2) 26 to 7 August 2020, producing 174 searches over 29 time points across this time frame.
A single reviewer screened titles and abstracts for relevance to the three dilemmas [19][20][21][22][23][24] , and reviewed full texts of relevant articles to extract data using a standard extraction form. Where full texts could not be obtained, data extraction was performed using the study abstract if possible. Screening results and data extraction were regularly cross-checked by three reviewers at intervals of 7-14 days between May and August 2020. Disagreements were resolved by consensus, with one reviewer acting as arbitrator if required. Data were extracted for study design, major themes and statements that updated the surgical literature from the rapid reviews developed in April 2020 [19][20][21][22][23][24] . Extracted data from studies with design ranking above 'poor' for all dimensions within Evans' hierarchy of evidence 28 , or those providing novel data to the surgical literature, were favoured in data synthesis. Following data extraction and thematic analysis of extracted narrative data, four reviewers inspected the authors' initial evidence-based recommendations (developed in April 2020) [19][20][21][22][23][24] for accuracy as of August 2020.
Data were synthesized in narrative and tabular formats. Outcomes of interest were: the proportion of relevant studies of weaker or stronger design according to Evans' hierarchy of evidence 28 , proportion of each study design, major themes within the narrative data of the relevant studies, and consistency of narrative data in August 2020 with the April 2020 evidence-based recommendations [19][20][21][22][23][24] .

Conventional systematic review
Conventional systematic review was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines 29 . The study protocol was prospectively registered on PROSPERO (CRD42020205845). Studies were eligible for inclusion if they provided or summarized recommendations at the specialty level or more broadly for surgical practice during the COVID-19 pandemic, and outlined a search strategy in the article or abstract that identified at least one database or search term. Studies of any design, in any setting, were searched for within PubMed (incorporating MEDLINE) and Embase using the search terms (surgery* OR surgical* OR surgeon*) AND ('COVID-19' OR coronavirus OR 'SARS-CoV-2' OR '2019-nCoV' OR 'corona virus' OR 'COVID'). This search was repeated six times, at monthly intervals, between 31 December 2019 and 30 June 2020.
Two reviewers independently screened titles and abstracts, reviewed full texts and extracted data using a standard extraction form. Screening of titles and abstracts was facilitated through use of a web application (Rayyan, Qatar Computing Research Institute, Ar-Rayyan, Qatar) 30 . Disagreements were resolved by consensus, with a third reviewer acting as arbitrator if required. Data were extracted for study design, setting, surgical specialty or area, methodological quality information, major themes, and statements providing recommendations for surgical practice during COVID-19.
Data were synthesized from all included studies in both narrative and tabular formats. The search did not identify any original research eligible for inclusion, so the likelihood of publication bias could not be assessed. The primary outcome was methodological quality of the included studies, which was independently assessed by two reviewers using the Assessment of Multiple Systematic Reviews (AMSTAR) tool 31 for included systematic reviews, and the Appraisal of Guidelines, Research and Evaluation (AGREE II) tool 32 for all other included studies. The secondary outcome was dissemination of data within the peerreviewed literature, which was inferred from individual numbers of citations according to targeted searches of Scopus (until 27 August 2020). For any studies unable to be identified on Scopus, numbers of citations were obtained through targeted searches of Google Scholar (until 27 August 2020). Four reviewers evaluated thematic evolution in extracted narrative data. Meta-analysis was not conducted due to heterogeneity within the included studies.

Results
Via 174 targeted searches staggered at approximately weekly intervals over 29 time points between 31 December 2019 and 7 August 2020, a total of 5238 records (4116 unique reports) were identified, of which 1256 were included in the rapid reviews ( Table  S1). The six systematic searches, at six time points at monthly intervals between 31 December 2019 and 30 June 2020, identified a total of 6385 records (5530 unique reports). From this total, 318 full-text articles were retrieved, of which 21 studies were included in the systematic review ( Fig. 1 and Figs S1-S6). Two full texts could not be obtained for evaluation, despite eligibility being identified on screening of title and abstract 33,34 .

Rapid reviews
The weekly volume of studies relevant to the surgical areas of interest [19][20][21][22][23][24] increased with time during the pandemic, peaking between 22 July and 31 July, when 143 studies were retrieved (Fig.  2). After the publication of the authors' initial recommendations 20,22,24 , 1017 relevant articles were retrieved. Of the relevant studies within the rapid reviews, 903 (72 per cent) had a lowerquality design (Fig. S7) 28 , with 402 (32 per cent) being opinionbased evidence or letters ( Table 1).
The first relevant study, chronologically, classified as stronger evidence was an observational study identified in the final search of March 2020 35 . Observational studies were more prevalent later in the time frame, particularly in June 2020 (median 6 (i.q.r. 5-7) per cent), peaking in the final week of analysis with 11 per cent (Table S2 and Fig. S11). Retrospective data analyses were the most common forms of stronger evidence (median 4 (i.q.r. 0-7) per cent) ( Table 1, Fig. S12). However, these were only retrieved after 15 April, peaking between 1 July and 31 July 2020 (median 8 (i.q.r. 7-11) per cent) ( Table S2 and Fig. S13).
From the 174 targeted searches, 48 (4 per cent) evidence-based recommendations were retrieved (median 2 (i.q.r. 0-4) per cent), compared with 209 (17 per cent) expert consensus recommendations in the same time frame (median 16 (i.q.r. 11-28) per cent) (Tables S1 and S2). The earliest evidence-based recommendations were retrieved in the first week of April 2020. Expert consensus recommendations were retrieved at rates ranging from two to 13 times that of evidence-based recommendations between 1 April and 31 July 2020 (Figs S14 and S15).

Consistency of surgical literature
Ongoing monitoring of narrative data between 31 December 2019 and 7 August 2020 identified no substantive thematic changes over time (Table S3). Despite substantial heterogeneity in study design, narrative data relevant to the three aspects of surgical care [19][20][21][22][23][24] consistently concurred with the authors' recommendations, apart from a small number of low-quality studies recommending an open approach over the laparoscopic alternative to minimize the risk of SARS-CoV-2 transmission to operating room staff 55 . As the pandemic progressed, thematic consistency was observed in all areas despite 1017 relevant publications. After accounting for inherent differences in specialty, there was no significant narrative contradiction in the evidence-based recommendations within the conventional systematic review.
Following consideration of the total body of surgical literature published between 31 December 2019 and 7 August 2020, inspection of the initial evidence-based recommendations made in the authors' three rapid reviews (developed in April 2020) 19,21,23 revealed no strong evidence subsequently published (as of 7 August 2020) to justify significant change in any recommendation (Table S4).

Discussion
This study has provided a comprehensive characterization of the surgical literature and its temporal evolution during the COVID-19 pandemic. Lower-quality studies predominated, with opinion-based articles and letters the most common communications. Quality did not improve over time despite the increasing number of relevant publications. Clinical recommendations were mostly based on expert consensus, opinion or narrative review.   There was a dearth of evidence-based guidance, which ranged from low to moderate methodological quality and for which numbers of citations reflected month of database entry. Evidence-based recommendations developed by the present authors in April 2020 remained clinically valid as of 7 August 2020 despite the large number of relevant publications during this intervening period. Although relevant studies of strong design have been published after the final search was completed 56 , these data have generally concurred with the present study. The initial evidence-based recommendations from April 2020 remain valid. Declaration of the COVID-19 pandemic 57 was followed by the proliferation of a large volume of related literature, demonstrating unprecedented responsiveness by peer-reviewed medical journals 25 . However, this did not result in corresponding benefit for clinical practice. Most studies were found to be of lower-quality design, and 4 months of further literature did not highlight significant modification to the present authors' early-stage, evidence-based recommendations 13 .
Rapid reviews with clearly specified research questions can produce similar clinical conclusions to systematic reviews, albeit with lower certainty 27 . The authors' April 2020 rapid reviews targeting specific clinical dilemmas 18 provided reliable and timely evidence-based solutions [19][20][21][22][23][24] .
The quality of surgical literature has been a topic of past controversy 58 . Although improved through consensus efforts valuing evidence-based principles 17 , the present study reveals that COVID-19 may have caused this issue to resurface. The predominance of expert opinion suggests that the global surgery community may have reverted to traditional eminence-based medicine 59 . Deviating from objective data risks differing clinical approaches being advised, potentially compromising safety 60 . Increasing ideological heterogeneity could confound attempts to achieve collective goals for global surgery 11 . A continued increase in global caseload 61 implies a substantial challenge may be created for the future.
The present study addressed multiple challenges for surgery during the pandemic 1 identifying clinical solutions that are relatively inexpensive and implementable on a global scale. Surgery on patients infected with SARS-CoV-2 may cause postoperative morbidity and mortality 3,4 , and this should be understood for preoperative risk-benefit assessment. Adequate delay after infection should be allowed to ensure safety for patients undergoing elective procedures 62,63 . Operations on COVID-19 patients should ideally occur in designated theatres with negative-pressure ventilation 12 , and suspected COVID-19 patients should wear a surgical mask during perioperative theatre transport 21 . Appropriate preoperative screening for active SARS-CoV-2 infection using patient history and reverse transcription-polymerase chain reaction is essential 5 . Triage should consider local COVID-19 prevalence and hospital resources 23 , and methods of ensuring a safe resumption of elective surgery despite high prevalence have been described 64 . Although caution was advised for laparoscopic surgery throughout the pandemic, the present study found no evidence that laparoscopy presented a greater risk to theatre staff than open surgery with respect to SARS-CoV-2 transmission. However, given theoretical aerosol transmission risk, capture devices should be used with energy sources producing surgical smoke 8 and desufflation of pneumoperitoneum should be performed using a suction irrigator system attached to a high-efficiency particulate air filter 19 . Given evidence of multisystem biodistribution 7 , all biological material should be treated as a potential SARS-CoV-2 source. PPE with aerosol precautions (N95 respirator or equivalent) 65 can provide adequate protection against SARS-CoV-2 66 , and is essential during surgical emergencies, aerosol-generating procedures or when a patient is not confirmed to be COVID-19 negative 21 .
The present study included the use of temporally staggered searches to capture subtle updates in a rapidly evolving evidence base potentially missed by searches at a single time point. Synthesis of a large amount of narrative data provided inexpensive potential clinical solutions. To increase worldwide applicability and reduce the risk of bias, searches were not limited by language 67 and globally accepted methods were used for classifying strength of evidence 28 , reporting methodology 29 and risk of bias assessment 31,32 . Despite these strengths, this study has limitations. The most significant of which was potential bias accompanying the selection of relevant studies for the rapid reviews. The rapid reviews targeted key dilemmas for pandemic surgery [19][20][21][22][23][24]   limited scope. The paucity of strong evidence in the reviews prompted caution regarding statistical bias within included studies. Heterogeneity in study design prevented meta-analysis, requiring data synthesis in multiple formats for maximal clinical utility and acknowledgement of likely biases.
Characterization of temporal changes in the surgical literature through a multifaceted systematic review confirmed correctness of an evidence-based approach tailored to surgical management during the COVID-19 pandemic. Synthesis of these results will provide surgical staff with an enhanced understanding of the trustworthiness and dissemination of the recommendations within the provided studies, that will allow them to decide whether they should adopt the recommendations for clinical care of surgical patients at their local healthcare institutions. Combination of findings from this study with the authors' experience during the COVID-19 pandemic provides an evidence-based model for rapid clinical change in surgical management ( Table 3).
The authors hope that these findings provide a crucial lesson for global surgery. Widespread adoption of evidence-based methodology targeting specific clinical dilemmas may facilitate the