SARS-CoV-2 vaccination modelling for safe surgery to save lives: data from an international prospective cohort study

1The SARS-CoV-2 pandemic has disrupted elective surgery globally, with millions of elective operations either postponed or cancelled. Restriction of elective surgery activity was necessary during periods of high COVID-19 hospital admissions in order to divert hospital resources to ICUs. However, in many countries, owing to safety concerns, reductions in COVID-19 admissions have been associated with only limited recovery of surgical services. Even in areas with low community SARS-CoV-2 infection rates, surgical patients are at risk of nosocomial SARS-CoV-2 infection, which is associated with high rates of postoperative pulmonary complications and death,. Although mitigation measures, such as preoperative reverse transcription (RT)–PCR swab screening and COVID-free surgical pathways, can reduce the risk of COVID-19 complications,, they are unlikely to be implemented universally, particularly in low- and middle-income countries (LMICs).

2Several SARS-CoV-2 vaccines have been authorized internationally following phase III trials that demonstrated 100 per cent effectiveness in preventing COVID-19-related deaths. Preoperative vaccination could support safe reinitiation of elective surgery by significantly reducing the risk of COVID-19 complications in patients undergoing elective procedures. However, it is projected that the population in only 37 countries will have widespread access to vaccination in 2021, with most LMICs only achieving widespread coverage from late 2022 onwards,. Consequently, throughout 2021, most governments will prioritize access to vaccination to patients at greatest risk of COVID-19 mortality.

3The benefits of SARS-CoV-2 vaccination in surgical patients are currently unknown, and therefore governments are not prioritizing them for vaccination. The aim of this study was to inform vaccination prioritization by modelling the impact of vaccination on mortality in patients undergoing any type of inpatient elective surgery.

4This study explored the impact of SARS-CoV-2 vaccination in adult patients (age at least 18 years) undergoing any type of elective inpatient surgery. The primary outcome was number needed to vaccinate (NNV) to prevent one COVID-19-related death over 1 year after SARS-CoV-2 vaccination, assuming that surgical patients would receive the vaccination before operation. When comparing NNV values for two different groups, the lower NNV value is more favourable, as this indicates that fewer people in that group need to be vaccinated in order to prevent one death. The secondary outcomes were NNV to prevent one COVID-19-related death over 30 days after SARS-CoV-2 vaccination, and additional COVID-19-related deaths prevented by prioritizing preoperative vaccination for surgical patients versus age-matched controls.

5As SARS-CoV-2 mortality is strongly associated with age, NNV estimates were preplanned to be stratified by age group (18–49 years, 50–69 years, 70 years or more). In addition, as prioritization and planning of cancer and non-cancer surgery differs, NNV estimates were also preplanned to be stratified by indication for surgery (cancer versus non-cancer). To contextualize the NNV estimates for surgical patients, age-stratified NNVs were also estimated for the general population. The study was registered at Clinical.Trials.gov (NCT0450998690).

6NNVs in surgical patients were based on postoperative SARS-CoV-2 rates and mortality from the multinational multicentre GlobalSurg–CovidSurg Week study. NNVs in the general population were based on global community SARS-CoV-2 incidence and UK SARS-CoV-2 case fatality data. It was assumed that, outside of the initial 30-day postoperative window, during the following 335 days surgical patients are at the same risk of SARS-CoV-2 infection and SARS-CoV-2 mortality as the general population within the same age group.

7The following equation was used to calculate NNV to prevent one COVID-19-related death over 1 year after SARS-CoV-2 vaccination in the surgical groups: where S is postoperative mortality (within 30 days after surgery) attributable to SARS-CoV-2 infection, D is the daily community SARS-CoV-2 incidence, and C is the case fatality rate for SARS-CoV-2 in the general population.

8To calculate the NNV to prevent one death over 1 year after vaccination in the general population, the equation was:

9Real-world estimates for postoperative SARS-CoV-2 rates and postoperative mortality attributable to SARS-CoV-2 were based on the GlobalSurg-CovidSurg Week study. This was an international prospective multicentre cohort study (NCT04509986) that included patients undergoing any type of surgical procedure performed routinely in an operating theatre by a surgeon. Participating centres collected data on all consecutive patients in one or more preselected surgical specialties during 1 or more weeks within an overall 28-day study window (5 October 2020 to 1 November 2020). Patients who had either elective surgery (planned admission to hospital) or emergency surgery (unplanned admission), as either an inpatient (surgery with a planned overnight stay) or day case, were included. This analysis was restricted to patients having elective inpatient surgery. Mortality status was determined by 30-day postoperative follow-up. Patients with data missing on age, sex, ASA physical status grade, indication for surgery, grade of surgery, SARS-CoV-2 infection status, or mortality were excluded from this analysis, to allow consistent denominators throughout calculations.

10The study collected only routine, anonymized data with no change to clinical care pathways. A secure online REDCap database was used for data collection. The study was registered at each participating hospital in line with applicable regulations. In the UK, the study was registered as either a clinical audit or service evaluation (registration reference at lead site, University Hospitals Birmingham NHS Foundation Trust: CARMS-16328).

11The proportion of patients who died within 30 days of surgery with death attributable to a postoperative SARS-CoV-2 infection was calculated as I × R, where I is the postoperative SARS-CoV-2 rate in the first 30 days after surgery, and R is the postoperative SARS-CoV-2 attributable mortality (an adjusted estimate for the difference in mortality between patients who did and did not have postoperative SARS-CoV-2 infection).

12In the GlobalSurg–CovidSurg Week study, postoperative SARS-CoV-2 was defined as any diagnosis of SARS-CoV-2 made after surgery, up to and including postoperative day 30. This included both patients with and without symptoms. SARS-CoV-2 could be diagnosed based on a positive RT–PCR swab, positive rapid antigen test, positive chest CT, and/or contemporaneous clinical diagnosis (in the absence of negative RT–PCR swab results). Patients who had a diagnosis of SARS-CoV-2 at any time before surgery were excluded from the denominator for calculating postoperative SARS-CoV-2 rates.

13Multiple studies have established that patients who become infected with SARS-CoV-2 after operation are at increased risk of postoperative death,. However, there are likely to be confounding factors, with higher postoperative mortality rates in SARS-CoV-2-infected patients partly attributable to baseline differences such as older age or co-morbidity. Therefore, an unadjusted risk difference for mortality between patients with and without postoperative SARS-CoV-2 infection is likely to overestimate mortality attributable to SARS-CoV-2.

14To reduce confounding, data for elective inpatient surgery from the GlobalSurg–CovidSurg Week study were used to estimate adjusted differences in mortality between patients who did and did not have postoperative SARS-CoV-2. This was based on average marginal effects taken from multilevel logistic regression models. The outcome in the models was 30-day mortality. As well as postoperative SARS-CoV-2 infection, the models were adjusted for factors that have previously been identified as independent predictors of death in patients with perioperative SARS-CoV-2 infection: age, sex, ASA physical status grade (grades I–II versus III–V), indication (surgery for benign versus malignant disease), and grade of surgery (minor versus major, based on the BUPA schedule of procedures). In addition, country was included as a random effect, with hospital nested within country. Models were stratified by age subgroup (20–49 years, 50–69 years, 70 years or more). Analyses were completed in Stata® version 15.1 (StataCorp, College Station, Texas, USA).

15Daily community SARS-CoV-2 incidence was based on SARS-CoV-2 cases reported globally in 2020. Countries were split into tertiles by community SARS-CoV-2 incidence rank. Median daily community SARS-CoV-2 incidence was calculated for each tertile to produce values for low, medium, and high community SARS-CoV-2 incidence. The main analysis was based on medium SARS-CoV-2 incidence.

16Age-stratified SARS-CoV-2 case fatality rates in the general population were calculated using SARS-CoV-2 seroprevalence and mortality data from England. The full methodology is described in Appendix S1.

17No deaths from COVID-19 have been reported beyond 1 week after SARS-CoV-2 vaccination in the phase III trials published to date. The main analysis was therefore based on SARS-CoV-2 vaccination being 95 per cent effective in preventing death from COVID-19.

18The number of additional COVID-19-related deaths that could be prevented in 1 year by vaccinating surgical patients before operation rather than the general population was calculated by subtracting the number of COVID-19-related deaths that would occur over 30 days in the unvaccinated general population from the number of COVID-19-related deaths that would occur in first 30 postoperative days in unvaccinated surgical patients. Estimates were age-stratified (Appendix S1).

19NNV values vary depending on local SARS-CoV-2 incidence. Therefore, a significant area of uncertainty concerns SARS-CoV-2 incidence in the general population, as this varies across both geographical regions and time. Future variation may reflect changes in government policies, or the emergence of new viral strains with either increased or decreased transmissibility. One-way sensitivity analyses were therefore undertaken, applying the low and high community SARS-CoV-2 incidence estimates.

20Best- and worst-case scenarios were produced to further explore uncertainty. The best-case scenario represents the lowest likely value for NNV and the highest likely value for additional COVID-19-related deaths prevented, whereas the worst-case scenario represents the highest likely value for NNV and the lowest likely value for additional COVID-19-related deaths prevented. The parameters for the best- and worst-case scenarios are described in Appendix S1 and summarized in Table 1.

21Overall, the GlobalSurg–CovidSurg Week study captured data for 141 582 patients from across 1667 hospitals in 116 countries. Of the 89 225 adults who underwent elective surgery, 31 434 had a day-case procedure and 1202 had a preoperative SARS-CoV-2 infection. Therefore, 56 589 patients were included in the main analyses (Fig. S1). A demographic breakdown is provided in Table 2.

22The overall postoperative SARS-CoV-2 rate was 0.96 per cent (541 of 56 589). Incidence by subgroup is shown in Table S1. Overall, 504 of 541 diagnoses (93.2 per cent) were based on a positive RT–PCR (495) or positive rapid antigen test (15). A further 19 patients (3.5 per cent) had a positive chest CT, and the remaining 18 (3.3 per cent) had a clinical diagnosis.

23In adjusted multilevel models, SARS-CoV-2 infection was significantly associated with 30-day postoperative mortality across all age groups (Table 3). Adjusted differences in 30-day mortality between patients with and without postoperative SARS-CoV-2 infection were calculated from these models and are presented in Table 4.

24In the main analysis, NNVs to prevent one COVID-19-related death over 1 year were lower for surgical patients than the general population in all age groups (Fig. 1). NNVs were lowest in people aged at least 70 years: 351 (196 for best-case, 816 for worst-case scenario) among those needing cancer surgery, 733 (407, 1664) for those needing non-cancer surgery, and 1840 (1196, 3066) for the general population. However, NNVs in patients aged 50–69 years needing either cancer surgery (559; 304, 1482) or non-cancer surgery (1621; 854, 4577) were favourable compared with the NNV for the general population aged 70 years or more.

25In sensitivity analyses modelling both low and high community SARS-CoV-2 infection rates, NNVs for surgical patients remained favourable (Table 5). Across all countries, there was an advantage to prioritizing surgical patients (Fig. 2). Relative to the general population, vaccination of surgical patients had the greatest advantage in countries with low community SARS-CoV-2 infection rates.

26Within each age group, NNVs to prevent one COVID-19-related death over 30 days were lower for surgical patients than for the general population, regardless of community SARS-CoV-2 incidence (Table 6). Across age groups, NNVs were lowest in people aged 70 years or more: 425 (231 for best-case, 1080 for worst-case scenario) for those needing cancer surgery, 1157 (592, 3316) for those needing non-cancer surgery, and 22 384 (14 552, 37 302) for the general population.

27Globally, a policy of preoperative vaccination of patients aged 70 years or more before elective surgery in preference to age-matched general populations was projected to prevent an additional 26 624 (9865 for worse-case, 50 410 for best-case scenario) COVID-19-related deaths in 1 year, assuming that global surgical activity was at 75 per cent of prepandemic volume (Table S7). Prioritizing all surgical patients for preoperative vaccination ahead of the general population was projected to prevent an additional 58 687 (20 177, 115 007) COVID-19-related deaths.

28This study used real-world data from an international, prospective cohort study to model NNVs to prevent COVID-19-related deaths among patients needing elective inpatient surgery. When compared within age groups, NNVs were consistently lower for surgical patients than for the general population. NNVs were particularly favourable for patients needing cancer surgery, who are likely to be prioritized for elective surgery as services restart in 2021. These findings were consistent across all settings regardless of community SARS-CoV-2 infection rates, as well as in best- and worst-case scenario analyses.

29SARS-CoV-2 vaccination could prevent tens of thousands of COVID-19-related postoperative deaths. However, vaccine supplies are likely to remain limited in most countries throughout 2021. Most governments are therefore prioritizing vaccination for groups at highest risk of COVID-19 mortality. This modelling study can inform prioritization of surgical patients within vaccination plans. It supports prioritization of patients aged 70 years or more needing elective surgery alongside other high-risk groups during early vaccination programmes. Once vaccines are rolled out to the wider population, it will be advantageous to prioritize surgical patients, particularly those undergoing cancer surgery.

30Implementation of vaccination for surgical patients will require the development of preoperative pathways to deliver vaccination ahead of planned surgery dates. These pathways should be designed alongside wider system developments aimed at reducing nosocomial SARS-COV-2 transmission, such as preoperative SARS-CoV-2 swab testing and COVID-free surgical pathways,. Future research should evaluate which of the licensed vaccines is most effective in surgical patients and the optimal timing for preoperative vaccination.

31There may be additional benefits to vaccinating surgical patients. Up to 70 per cent of elective procedures were postponed during the first wave of the SARS-CoV-2 pandemic, resulting in an estimated 28 million elective operations being delayed or cancelled,. Although elective surgery volumes have started to recover in many countries, ongoing disruption is likely to continue throughout 2021, particularly in the event of countries experiencing further SARS-CoV-2 waves. Prioritization of SARS-CoV-2 vaccination for surgical patients could support safe reinitiation of elective surgery services, especially in regions where vaccinating the total population will take several years. In addition, SARS-CoV-2 vaccination is likely to decrease postoperative pulmonary complications, reducing intensive care use and overall healthcare costs.

32This study has limitations. First, NNV estimates should be interpreted with caution. Although NNVs for surgical patients are advantageous compared with those for the general population in sensitivity analyses for both low and high SARS-CoV-2 infection rates, precise NNV values fluctuate depending on prevailing SARS-CoV-2 infection rates; the higher the SARS-CoV-2 incidence, the fewer people need to be vaccinated to prevent one SARS-CoV-2 infection, so the more favourable (lower) the NNV becomes. Second, NNV estimates were based on postoperative SARS-CoV-2 infection rates taken from a global snapshot in October 2020, and it is unknown how these rates vary over time and across regions. Third, SARS-CoV-2 case fatality rates were based on data from England and it is unknown how generalizable these are, although they are broadly consistent with other published data (Appendix S1). Finally, vaccine effectiveness was modelled based on the limited data available from trials performed in the general population and it is unknown how applicable this is to surgical patients.