Mini abstract
We tested usability, safety, and ability to manage elective surgery flow of a multidisciplinary pathway adopting the SWALIS-2020 model in an interhospital setting during the COVID-19 pandemic. This pilot included 295 adults. The pathway prioritized, monitored and scheduled surgery in an 840.000-inhabitants area. Using the COVID-19-GOA-Sur-MDT-SWALIS-2020 model may be warranted.
Structured abstract The COVID-19 outbreak burdens non-COVID elective surgery patients with figures similar to the SARS-Cov-2, by creating an overwhelming demand, increasing waiting times and costs. New tools are urgently needed to manage elective access. The study assesses the “SWALIS-2020” model’s ability to prioritize and optimize access to surgery during the pandemic.
A 2020 March - May feasibility-pilot study, tested a software-aided, inter-hospital, multidisciplinary pathway. All specialties patients in the Genoa Departments referred for urgent elective surgery were included in a multidisciplinary pathway adopting a modified Surgical Waiting List InfoSystem (SWALIS) cumulative prioritization method (PAT-2020) based on waiting time and clinical urgency, in three subcategories: A1-15 days (certain rapid disease progression), A2-21 days (probable progression), and A3-30 days (potential progression).
Following the feasibility study (N=55 patients), 240 referrals were evaluated in 4 weeks without major criticalities (M/F=73/167, Age=68.7±14.0). Waiting lists were prioritized and monitored, and theatres allocated based on demand. The SWALIS-2020 score (% of waited-against-maximum time) at operation was 88.7±45.2 at week 1 and then persistently over 100% (efficiency), over a controlled variation (equity), with a difference between A3 (153.29±103.52) vs. A1 (97.24 ± 107.93) (p <0.001), and A3 vs. A2 (88.05±77.51) (p<0.001). 222 patients underwent surgery, without related complications or delayed/failed discharges.
The pathway has selected the very few patients with the greatest need, optimizing access even with +30% capacity weekly modifications. We will use the pathway to manage active, backlog, and hidden waiting lists throughout the further pandemic phases, and are looking for collaboration for multi-center research. https://www.isrctn.com/ISRCTN11384058.
Background
Nations all over the world are currently under healthcare pressure due to the Coronavirus 2 (COVID-19) pandemic. Most of the hospitals have become points of care for COVID-19 patients, with temporary capacity changes involving anesthetists and theatre staff, who were diverted to Intensive Therapy Units (ITUs) managing severe acute respiratory syndrome from COVID-19 (SARS-Cov-2). An immediate consequence of these re-arrangements is the reduction in surgical capacity, forcing hospitals to run only emergency theatres and suspend nearly all elective complex and cancer operations1–3, as well as most of outpatient care. Knowledge of the problem has reached the scientific4 and wider communities5, disclosing figures of patients at risk of a similar magnitude to that directly caused by the viral infection6, and summing COVID-19 and cancer burdens7. Services face the challenge to prioritize access to surgery, streamlining patient flow and managing an overwhelmingly accumulated demand for surgery, expected to nearly triplicate in the next year, with massive additional costs8. However, not all progressive diseases determine the same urgency. In cancer patients, tumor biology might be categorized on its speed. Several national professional efforts have delivered guidance to select patients for cancer surgery in the COVID-19 era, establishing their level of urgency9–12, with specific proposals13. The problem is growing further in complexity, costs and time, far surpassing initial figures, due to the elective surgery backlog and “hidden” waiting lists14, determining an unprecedented, overwhelming and long-lasting demand/supply imbalance15.
In Liguria, the metropolitan area of Genoa (840.000 inhabitants), 2,000 hospital beds are provided by an urban health and social care Trust, several hospitals, and an adult research hospital (Policlinico San Martino - PSMRH), which is also a comprehensive cancer center. The normal total adult ITU capacity is 55 beds, in less than a month expanded to 113 (+100%) to cover the SARS-Cov-2 needs. Routine surgery has been suspended. PSMRH has been able to serve as a residual central hub, with most surgical services available. Out of the peak 72 ITU beds available at PSMRH, 25 (34%) were saved for non-COVID-19 care. Thanks to the 29 residual weekly theatre sessions, a small number of surgical patients could still be admitted in wards, where screening oropharyngeal swabs and chest x-rays were performed 24 hours to operation. At the beginning of March 2020, the Liguria Regional Health Trust (ALiSa) and the Regional Inter-Trust Surgical Departments have commissioned PSMRH to develop a collaborative centralized pathway for urgent high-complexity non-deferrable cases at PSMRH.
The recent review by Dery et. al. shows that patient prioritization could help to manage access to care in an equitably and transparently, however practical implementation is a challenge16.
One of the authors (RV) developed a prioritization method for surgical waiting lists (PAT IT 2007), later included in Italian National Research projects, leading to the development of the Surgical Waiting Lists (SWALIS) model to manage and prioritize access to elective surgery in 200917 (see Methods section), designed on implicit criteria (independent from specific diagnosis). Similar criteria are used in Italy18, Australia19, and the UK20. The prioritization model has shown the potential to protect patients from excessive waiting times, increase health gain21, selecting patients with equity17, to optimize service efficiency in theatre allocation22,23.
Aims
We report the feasibility-pilot of a novel, software-aided, inter-hospital surgical prioritized pathway covering the metropolitan area of Genoa (840,000 inhabitants) for all specialties in the Departments of Surgery, based on the new “SWALIS 2020” model to prioritize, monitor and optimize access to surgery during the COVID-19 pandemic.
Methods
Design
In mid-March 2020, a dedicated multidisciplinary committee (COVID-19-GOA-Surgery MDT) was formed by the Regional Inter-Trust Surgical Department heads, the heads of the PSMRH Department of Surgery, Cancer Board, Medical and Quality Care Unit, and Anesthesia Department, the referring surgical units’ representatives. The committee designed a bespoke policy (COVID-19-GOA-Surgery Policy) to accommodate surgery for un-deferrable patients, consistent with the existing governance. Per the Italian National urgency categories, patients are suitable for an urgent referral if they have been diagnosed with a worsening condition, with the risk of rapid deterioration18.
As no change was proposed in patients’ clinical treatment and in the existing governance, our study was set with as a single-cohort before-after service improvement research design, reported using the STROBE statement24. The study is registered as ISRCTN11384058. We started working in the third week of March 2020, designing our project in two phases:
Feasibility (March 31st - April 7th). Assessing the feasibility of a collaborative Service Improvement, including all hospitals in the Metropolitan area of Genoa into a temporary, centralized, pooled service.
Pilot (four weeks from April 8th – May 5th). Running the model and monitoring its impact, prioritizing and streamlining access to surgery.
Feasibility assessment
We initially assessed:
The governance issues and our developments’ compliance with the existing National and Regional directives.
The applicability of the prioritization model to the procedural principles set by the Liguria Health Authority commissioners and the PSMRH.
The willingness of clinicians to recruit patients.
The number and characteristics of eligible patients.
The subsequent pilot study outcome measures, based on the SWALIS performance indexes, and any further ones appropriate.
The measures of centrality and variation of these main outcome measures.
Availability of the data needed and development adequacy of software/database.
The overall time needed to manage the data.
The SWALIS 2020 Model
To prioritize the waiting list, we have adopted a modified SWALIS model (SWALIS 2020), as follows:
Clinical urgency assessment, following the Italian National urgency categories18, with specific adaptations grading the likelihood of progressing to deterioration or emergency for urgent cases. We have re-defined the model introducing three urgent subcategories: A1-15 days (certain rapid progression), A2-21 days (probable rapid progression), and A3-30 days (potential rapid progression) (Table 2).
A maximum waiting time is set for each urgency category (A1-15 days, A2-21 days, A3-30 days, B-60 days, C-180 days, D-360 days), as a key criterion for a time-based prioritization.
The list is dynamically ordered by computing a “priority” score (SWALIS 2020 score) for each referral, based on the waited time in relation to the corresponding maximum allowed (Figure 2a). All patients reach the top of the list at the speed set by their clinical urgency, progressing through pre-admission stages by a score obtained following a cumulative linear prioritization method (Patent pending, Figure 2b). Defining updated urgency (U0, U1, U2, … Un) at time of re-evaluation (t0, t1, t2, … tn), and defining P(tn+1 + U)= 1 as an expression of 100% of the maximum allowed waiting time, the cumulative priority (P) at the time of prioritization P(t) is defined as follows:
During the study, project participants were aware of the original SWALIS model general linear prioritization method (Figure 2a), while the new algorithm above (Figure 2b) was software-run and only known by the two inventors (SD, RV) as pre-existing intellectual property, under a Non-Disclosure-Agreement.
The waiting list is measured through purposely designed performance indexes21, including cross-sectional and retrospective views (Table 3).
The referral pathway
We have designed a new pathway based on the pooled referral, waiting list prioritization, admission schedule, and surgery at PSMRH facilities. It is shown in Figure 1.
Referral. Each Unit in the Metropolitan area selects amongst their waiting patients those in urgent category (< 30 days) and refers them to the pooled COVID-19-GOA-Surgery waiting list, the A-urgency category described above. The key information provided in the referral forms is summarized in Appendix 1. This includes clinical and socio-organizational information, i.e. expected procedure time and complexity, ITU and total stay, and socio-familial discharge complexity. The referring surgeons also acknowledged that in the context of the COVID-19 pandemic, the value of theatre time is extremely precious, that failure to rescue is increased and patients run the additional risk of hospital-acquired SARS-COVID-19 infection.
MDT clinical triage. All new referrals are assessed once weekly for appropriateness by the COVID-19-GOA-Surgery MDT in a videoconference. Appropriateness includes governance, clinical, and organizational aspects. According to the Regional Health Service policies, in order to maximize safety and quality of care, referrals from general surgery units are addressed to super-specialty hubs at PSMRH.
Priority based scheduling
the “SWALIS 2020” method. The list is ordered dynamically on the basis of the SWALIS 2020 score (i.e. the percentage of patients’ maximum allowed waiting time), according to a novel method (RV, SD) adopting linear cumulative prioritization, including recording any change in the level of urgency (Figures 1, 2)
Admission and theatre scheduling. Whilst exploring the forthcoming calendar, the waiting list order is computed by the future theatre scheduling date. Some degree of practical flexibility is allowed to each Surgical Unit, in scheduling their surgeries preventing the waste of theatre time, provided patients close to breaching their maximum time are scheduled. In case of a last-minute re-schedule or cancellation, the same Unit schedules the next closest suitable patient on the priority list.
In order to allow for COVID-19 swab analysis, patients are scheduled for admission 48 hours prior to surgery based on theatre availability, expected ITU needs, length of stay, and complications of discharge.
Post-discharge. The pathway also includes admission, post-discharge, and follow-up arrangements.
Data management
During both the 1) feasibility and 2) pilot phases, we have refined the referral process along with its application, discussing each change at the COVID-19-GOA-Surgery MDT. Data was admin-checked at referral for completeness, and at MDT for appropriateness. During the feasibility phase we have progressively strengthened the data-system tools, upscaling from 64-bit password-encrypted email referral archives, password-encrypted shared secured folders, spreadsheets (MS Excel™), and database live-running user interface (MS Access™) code-developed on Visual Basic for Application (MS VBA™).
Statistical analysis
Continuous variables are reported as means (SDs), median or 25th-75th percentile, while categorical variables as frequencies and percentages. The Shapiro-Wilk test was used to assess the normal distribution of the continuous variables both in the whole patient population and in each urgency category. When the Shapiro-Wilk test reaches statistical significance, the distribution is non-normal. Continuous variables were compared by using the Mann-Whitney U test or one-way ANOVA with the Kruskal-Wallis rank-sum test. For the ANOVA models that reached the overall significance, the Dwass-Steel-Critchlow-Fligner method was used for pairwise comparisons. Continuous variables were also evaluated with Spearman’s rank for correlation analysis. Loess (locally weighted smoothing) was used as non-parametric smoother for regression analysis. Most of the statistical analyses are provided in the electronic supplementary material (ESM). Statistical significance was assumed with a two-tailed p value of less than 0.05. Statistical analyses were performed with the R statistical environment (version 3.6.3).
Results
1. Feasibility
The SWALIS model run immediately after commission, while assessing the full feasibility. The results are summarized below.
a) Information governance required 64-bit cryptography. The Regional Ethic Committee approved the project waiving the patient consent, given no change was made to the current clinical practice.
b) COVID-19-GOA-Surgery MDT confirmed the applicability of the SWALIS 2020 model;
c) All referring clinicians confirmed their willingness to recruit all eligible patients.
d) The two-week demand counted 55 patients from 9 specialties, against a capacity for approximately 30 high-complexity surgeries weekly. The available service initially counted 29 weekly theatre sessions (3.4 /100,000 inhabitants) as Breast, Colorectal, Endocrine, Ear-Nose-Throat, Hepatopancreatobiliary, Maxillo Facial, Plastic, Sarcoma and Melanoma, Esophagogastric, Urology, and Functional General Surgery, making a reduction to one-sixth weekly sessions.
e) A semi-quantitative definition of clinical urgency was proposed (Table 1), all patients were discussed at the MDT videoconferences to achieve full consensus prior to scheduling. Overall, 55 patients were referred. The SWALIS 2020 outcome indexes included the SWALIS 2020 score at admission and deviation events, the number of postponements (prior to admission) and cancellations (on the day) (Table 2).
View this table:View this table:f) The results at weeks 1 and 2 are shown in Table 3. Waiting lists consisted of 51 and 60 patients respectively, with admission priority scores of 88.7% (±45.2) and 118% (±60.1).
View this table:g) The procedure and database were defined; referrals were sent with approximate 70% overall completeness; the run-time prototype was utilized in the MDT videoconferences.
h) At meetings end data was 100% complete, clearing any missing data in run-time.
2. Pilot
Safety
The full results are shown in Table 3. Two-hundred-forty patient referrals (Age=68.7±14.0, Female/Male=73/167) were recorded in four weeks with no reported major criticality from 12 Surgical Units, and 221 patients underwent surgery (Age=67.7±14.8, Female/Male=73/148). No major adverse event has occurred. No perioperative complication, hospital stay prolongation, or failed discharge was reported as caused by the allocation model.
Appropriateness
Two hundred and ninety-five patients have been assessed by the COVID-19-GOA-Surgery MDT. The COVID-19-GOA-Surgery MDT participants agreed after changes from the referrals in about 20% of cases, reassessing clinical urgency (n=32), expected operation length (n=21), surgical indication (n=4). Five (1.7%) patients were excluded following MDT. Eleven referrals have been placed against regional and health directives, triggering governance responses.
Service performance
Charted data are shown in Figure 3. Two-hundred-twenty-one patients have been scheduled as of 9 May 2020. Seven A1-urgency-patients could be scheduled in the week immediately following discussion due to major worsening conditions (progressive upper airway obstruction, near-irresectability or severe anemia).
At week 1 167% of the patients discussed exceeded theatre capacity. The length of the waiting list constantly increased until week 6, when an increase in theatre availability allowed for more adequate processing. The model has since made it possible to constantly monitor for progressive priority score increase (clinical safety and effectiveness) allowing for scheduling of approximately 60% of highest priority patients, while monitoring priority ranges (vertical and horizontal equity). About 40% of patients were admitted with a priority lower than the average, due to organizational reasons (last-minute reschedules, theatre time optimization, ICU capacity constraints, surgical team reasons), and never below 75%.
Deviations from planned surgeries occurred in 22 cases: positive swab (n=8), unavailable ICU bed (n=7), changed clinical conditions (n=6), and patient inability to reach the hospital (n=1).
Pathway Criticalities
We recorded the initial difficulty in filling the full referral form by approximately 20% of the surgeons, and an initial reluctance to collaborate in pooling the waiting list (3 out of 14 Units). The urgency category assignment has required discussion to reach homogeneity in approximately 5% of referrals. As the patient flow started to increase due to re-expanding theatre capacity, more flexibility of the priority-based allocation policy was required. Lastly, two referring Units have detoured 47 patients from the public hospital pathway, utilizing existing private conventions/contracts of individual surgeons, making maximum use of the Metropolitan area bed capacity.
Discussion
Given the baseline imbalance between supply and demand for surgery, the massive reduction in available resources is an acute worldwide issue in several specialties 5,12,20, causing great public concern5. This problem is evident in both the general district and central academic hospitals, regarding active, backlog, and hidden waiting lists; hence thousands of patients face the risk of not receiving timely or curative surgery5,14,25.
Since its debut in 2009, we are the first to report the use of the SWALIS model in an inter-hospital referral context. Our results might help the surgical community make the best use of the scarce theatre capacity left available by the COVID-19 pandemic, throughout its phases. For four weeks the COVID-19-GOA-Surgery has utilized the pathway to select, schedule and perform 221 major surgical procedures, transparently sharing indications for surgery among a wide workgroup, ensuring ethical and inclusive principles.
The assignment of the urgency category is a potential area of disagreement, but eventually all cases have easily been resolved at the weekly MDT meetings. The frequent urgency category re-evaluations of backlog patients have also benefited by the introduction of the new cumulative linear prioritization method to best log the patient history. The pathway has proven able to prioritize the demand, address its appropriateness, and live-monitor its changing (im)balance allocating the inconstant capacity in near-real-time. Patients could be informed of their expected waiting time. We initially measured an additional capacity need of approximately 87%. We have recorded outliers in priority on the waiting list (backlog referrals), and among the admitted patients, mainly caused by last-minute schedules to achieve theatre optimization. Approximatively 40% of patients have been operated on with low SWALIS 2020 priority scores. These patients were selected for organizational reasons, to achieve an efficient allocation of theatre capacity. While clinical effectiveness has not been tested directly, accepting the waiting time breach as a surrogate marker its control has proven positive too.
This is a preliminary pilot report in an extremely unusual condition, hence carrying limitations in internal validity, forced by the pressing need for an immediate clinical implementation in the unprecedented and complex context of the COVID-19 pandemic. We have initially focused on the waiting list prioritization and allowed free final patient scheduling, as such a policy was not the focus of our pilot. Additionally, our experience is limited to a small geographical area, and we have scheduled a small number of patients overall. Nonetheless, the COVID-19-GOA-Surgery MDT pathway is now established has smoothly processed dozens of patients affected by very wide diagnosis groups.
There are limits of external validity too, as this model has never been tested in a controlled experimental study, on a macro-regional scale. However, our experience is multi-institutional, and this early pilot has shown the COVID-19-GOA-Surgery-MDT-SWALIS-2020 pathway to be adaptable to last-minute changes of service demand or supply capacity with fairness, effectiveness, efficiency and transparency, by what has proven to be an easily scalable model.
The English National Health Service and Royal Colleges, American College of Surgeons, and the Emergency Surgery Associations have edited guidance and directives on COVID-19 related practice, on the selection of non-deferrable patients for urgent surgery11 and on the recovery of services26. As reported by Qadan et al., exemplary multi-site efforts have been made to maintain the necessary multidisciplinary approach for surgical cancer care27 and to clinically prioritize cancer surgery patients13, as proposed by Mazzaferro and colleagues. Our model is in full agreement and works complementarily with those.
There is intrinsic complexity in applying pooled-scheduling to a large multi-specialty inter-hospital environment. Any solution should combine rigorous principles with flexibility and applicability. At the UCSF, a prioritization schema has been utilized through a computerized dashboard to schedule surgery, successfully implementing collaboration, transparency, and operational efficiency28. Due to the simplicity of the time-based prioritization, our model is complementing those experiences too, by adding the implicit clinical urgency criteria, making it able to include a pooled multi-specialty, benign and cancer waiting list.
Conclusions
Our model complements the existing international experience and guidance on prioritization and recovery of surgical services. It has succeeded creating a collaborative inter-hospital pooled waiting list, accurately selecting the fewer cancer patients in most need to receive effective surgery, even with extremely scarce resources. It has been able to keep providing appropriate and audited access to surgery even in the case of rapid capacity modifications, planning, allocation and scheduling of theatre and intensive care resources, and managing backlog consistently. With adaptation to the normalizing patient flow, these findings encourage its wider use in the next phases of the pandemic which is is expected to bring about dramatic consequences on non-COVID urgent cancer and routine patients. The potential benefits of our model might be extended to all these, as a legacy model for a finely sustainable elective surgery patient flow, in surgical departments and hub-spoke networks.
Future steps & collaborations
We will continue using the COVID-19-GOA-Surgery-SWALIS-2020 waiting list prioritization model. We are looking for collaboration to design a prospective anonymized / network study testing a new priority-based allocation policy. We are available to share the key software modules (Patent pending) under Creative Commons (BY-NC-SA)6 license to hospitals for non-profit use.
Data Availability
The datasets generated during and/or analyzed during the current study are/will be available upon request from Mr Roberto Valente (Roberto.valente@hsanmartino.it) and Dr Stefano Di Domenico (Stefano.didomenico@hsanmartino.it). The dataset is in an MS Access TM format, and can be anonymized to any level required, as allowed by the Liguria Ethics Committee. The dataset will be made available 3 months after study completion, for 12 months, extendable. Any sharing request will be submitted to the Liguria Ethics Committee for approval.
Contributors
Roberto Valente has conceived, designed and coordinated the study, created the dataset and co-perfomed data extraction and analysis, oversaw the data collection and analysis, edited and proofread the manuscript. Stefano Di Domenico has co-designed the study, provided intellectual content, co-managed the dataset, oversaw the data collection and addressed its analysis, co-edited and proofread the manuscript. Matteo Mascherini has managed the dataset, reviewed and proofread the manuscript. Gregorio Santori has performed the data analysis, edited the charts, reviewed and proofread the manuscript. Francesco Papadia, Giovanni Orengo, Angelo Gratarola have provided intellectual content in designing the study and reviewed and proofread the manuscript. Ferdinando Cafiero and Franco De Cian have contributed as senior authors with relevant intellectual content, and reviewed and proofread the manuscript, taking full responsibility for the study.
Declaration of interests
None of the authors has any competing or conflicting interests with the submitted study.
Role of the funding source
The study has not received any external or ad hoc funding. Alisa contributed to maintaining the coherence of the study with institutional directives, and finally reviewing the manuscript. The corresponding author (RV) confirms that he had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Data sharing
The datasets generated during and/or analyzed during the current study are/will be available upon request from Mr Roberto Valente (Roberto.valente{at}hsanmartino.it) and Dr Stefano Di Domenico (Stefano.didomenico{at}hsanmartino.it). The dataset is in an MS Access TM format, and can be anonymized to any level required, as allowed by the Liguria Ethics Committee. The dataset will be made available 3 months after study completion, for 12 months, extendable. Any sharing request will be submitted to the Liguria Ethics Committee for approval.
2a) The linear prioritization method (Pat 2007, SWALIS 2009)
The referring surgeon declares patient’s clock start date (t0) and clinical urgency category (U) based on the likelihood of quick deterioration, to the point where it may become an emergency. Clinical urgency (U) is then associated to maximum waiting time from t0. In the SWALIS 2020 model U can assume six different values in days: U = {A1=15, A2=21, A3=30, B=60, C=180, D=360}.
Given on U and t0, and defining P(t0+U) = 1, the priority (P) at the time of prioritization P(t) is defined as follows: t10 = patient 1 clock start date; U1 = patient 1 urgency category maximum allowed waiting time; t10 = patient 2 clock start date; U2 = patient 2 urgency category maximum allowed waiting time; P1 = patient 1 priority at time of prioritization (t); P2 = patient 2 priority at time of prioritization (t).
2b) The cumulative linear prioritization method (Pat 2020, SWALIS 2020)
Clinical conditions can change during the waiting time (t0, t1, t2, … tn) affecting the patient’s urgency (U0, U1, U2, …Un). Priority can be calculated as summation, based on urgency variations: t0 = start waiting time; U0 = urgency for patient at starting time t0; tn = updated urgency time; Un = updated urgency for patient; t = time of prioritization.
The SWALIS 2020 prioritization method assumes four priority scores stages: “Ideal” (0-50%), color code white, “optimal” (51-75%) color code green, “due” (76-100%) color code yellow, “overdue” (>100%) color code red.
Acknowledgments
Prof. Angela Testi and Dr. Alessia Saverino gave insightful input in editing the manuscript. Ms. Marina Caldano provided data input and participated in the multidisciplinary meetings. Prof. Susan Marie Campbell was involved as a medical writer in the linguistic control of the manuscript.