The results of cardiac surgery during the COVID-19 pandemic compared with previous years: a propensity weighted study of outcomes at six months

Abstract

Objectives

In addition to excess mortality due to COVID-19, the pandemic has been characterised by excess mortality due to non-COVID diagnoses and consistent reports of patients delaying seeking medical treatment. This study seeks to compare the outcomes of cardiac surgery during and before the COVID-19 pandemic.

Design

Our institutional database was interrogated retrospectively to identify all patients undergoing one of three index procedures during the first six months of the pandemic and the corresponding epochs of the previous five years.

Setting

A regional cardiothoracic centre.

Participants

All patients undergoing surgery during weeks #13-37, 2015-2020.

Main outcome measures

Propensity score weighted analysis was employed to compare the incidence of major complications (stroke, renal failure, re-ventilation), 30-day mortality, six month survival and length of hospital stay between the two groups.

Results

There was no difference in 30-day mortality (HR = 0.76 [95% CI 0.27-2.20], p = 0.6211), 6-month survival (HR = 0.94 [95% CI 0.44-2.01], p = 0.8809) and duration of stay (SHR = 1.00 (95% CI 0.90-1.12), p = 0.959) between the two eras. There were no differences in the incidence of major complications (weighted chi-square test: renal failure: p = 0.923, stroke: p = 0.991, new respiratory failure: p = 0.856).

Conclusions

Cardiac surgery is as safe now as in the previous five years. Concerns over the transmission of COVID-19 in hospital are understandable but patients should be encouraged not to delay seeking medical attention. All involved in healthcare and the wider public should be reassured by these findings.

Introduction

We have previously reported our experience of maintaining a regional cardiac surgical service during the first surge of the COVID-19 pandemic in the UK. ¹ At the time of writing, the number of confirmed cases of COVID-19 worldwide exceeds 270 million, while in the UK the number of confirmed cases has surpassed 11 million and there have been more than 140,000 deaths. ²

On 29 April 2020, Sir Simon Stevens, the Chief Executive of the National Health Service (NHS), wrote to all NHS hospital CEO’s setting out requirements for managing ‘the second phase of the NHS response to the COVID-19 outbreak’. The recommendations and requirements reflect the continuing need for a substantial proportion of NHS resources to be deployed for the care of patients already infected as well as for additional capacity to be immediately available in the event of further surges. The letter referenced the sharp drop in emergency room attendance, predicting a rebound in the demand for acute cardiac surgery and cardiology, among other specialties, and specifically asks that all NHS hospitals ensure that ‘urgent and time-critical surgery and non-surgical procedures be provided at pre-COVID levels of capacity’; the letter also makes clear that the COVID-19 epidemic ‘is going to be with us for some time to come’. ³

Since that letter was written, the UK has experienced three further surges (Q3/4 2020, Q1 2021 and Q3 2021). This has not only required that we continuously adapt protocols for protecting both staff and patients but has resulted in further capacity constraints, leading to significant delays for almost all elective surgery and a considerable proportion of the urgent surgical workload. The undoubted success of the UK’s immunisation programme notwithstanding, further surges are to be expected while the NHS attempts to deal with the growing backlog of elective work. While capacity has been constrained because of the need to divert resources to deal with COVID-19 admissions, there has also been a decline in acute referrals and an understandable reluctance on the part of patients to be admitted to hospital for elective surgery, opting to remain at home on ever-growing waiting lists over being admitted to hospital with the attendant risk of catching COVID-19.

In this paper we present the outcomes of cardiac surgery in the first six months of the pandemic, compared with the outcomes of comparable cases in the corresponding epochs of previous years. In this way, we hope to be able to reassure both patients and the profession that cardiac surgery can be safely performed during the pandemic and that patients should not delay seeking medical attention because of it.

Materials and methods

The study was initiated as a part of our institution’s outcomes audit and was therefore exempt from the requirement for ethical committee approval.

As the pandemic has progressed, the regulatory body (NHS England) has updated the infection control requirements for all hospital admissions. ⁴ In summary, there has been a requirement to stream patients through green (low risk), amber (intermediate risk) and red (high risk) pathways for all procedures. Our infection control protocols have therefore been modified en passant to take account of guidelines as they emerged. Where clinically permissible, all patients were required to have had two negative COVID-19 polymerase chain reaction (PCR) swabs before undergoing surgery. Patients in whom this was not possible, who were transferred to our care as in-patients from other hospitals or were already known to be COVID-19 positive, were managed along amber/red pathways. Surgery for this group was deferred until they were confirmed to be COVID-19 negative, unless their clinical condition mandated urgent or emergency surgery sooner.

Patients and data collection

In the UK, outcome data on all patients undergoing cardiac surgery are submitted to the National Institute for Cardiovascular Outcomes Research (NICOR); all units are required to maintain a database for the purpose with dedicated staff responsible for the collection, completeness and submission of the data. Our institutional database was interrogated and the data entries for all patients undergoing cardiac surgery in the period 16 March–13 September 2020 (weeks 12–37) extracted. From this group, we extracted a cohort of patients undergoing one of three so-called index procedures: aortic valve replacement (AVR), coronary artery bypass grafts (CABG) or AVR&CABG combined.

The UK Office for National Statistics (ONS) publishes national mortality data weekly, comparing the number of deaths with a rolling average of the corresponding week in the previous five years. By way of a control group, therefore, we have compared the outcomes in these patients with a cohort(s) of patients operated upon during the corresponding epochs of each of the five previous years (weeks 12–37). Operative risk was estimated using the EuroSCORE II ⁵ algorithm, which has been extensively validated, is used throughout Europe and a summary of which is shown in supplementary Appendix I. The outcomes of interest are the incidence of major complications, postoperative stay, 30-day, and six-month survival.

Patients were not involved in the conception and design of the study, which was conducted retrospectively. However, we consider that the study contains information with important implications for patients contemplating surgery during the COVID-19 pandemic. We propose including a synopsis of our findings as part of our preoperative patient information and will gladly make our data and findings to other units.

Statistical method

Propensity score weighting was used to ensure that the comparison and selection assignment differences between the groups (pre-COVID 19 epochs and COVID-19 era) was balanced on observed and unobserved baseline covariates ⁶ and to control for selection bias. A propensity score was calculated for each patient based on two factors: EuroSCORE II and the index procedure. Patients who did not have data for these two factors were excluded. Inverse probability weighting, using average treatment effect (ATE) weights, generated by the propensity scores, ⁶ were used in the final analysis to compare outcomes between patients operated on during the COVID-19 era and in the pre-COVID 19 era. For the patients in the COVID-19 era group (2020) probability weights equal to 1/Propensity Score (PS) were calculated, and for those in the control group (pre-COVID 19 [2015–2019[) the probability weights were calculated as 1/(1-PS). These weights were then used in the analysis to “imitate” a control group (pre-COVID 19) which was comparable in terms of EuroSCORE and the procedure type to the COVID-19 era patients.

Differences in the proportions of major complications were calculated using a weighted chi-square test, presenting the p-value corresponding to the corrected weighted Pearson chi-square test statistic (F-statistic). Postoperative stay was analysed as a time to event outcome, from date of operation to date of discharge from hospital. A competing-risk survival regression model was used to incorporate death in hospital as a competing event. This model is based on Fine and Gray’s proportional subhazards model and is presented by cumulative incidence function curves, which indicate the probability of a patient being discharged from hospital before a given time. Patients were right-censored if they were missing date of discharge and were known to be alive. Survival time was calculated as the number of months from date of operation until date of death, or date of data extraction (25 January 2021), whichever was the sooner. Patients were right-censored if they were still alive by the date of data extraction. Survival is presented by Kaplan-Meier curves and compared between the two groups of patients by a weighted Cox-based test for equality. The proportional hazards assumption was tested by scaled Schoenfeld residuals, and by inspection of the hazard ratio plots. ⁷

The populations for analysis include the cohort of patients undergoing one of three index procedures (AVR, CABG and AVR&CABG) for both the COVID-19 era and pre-COVID-19 epochs. Descriptive data are expressed as mean ± standard deviation or median [interquartile range].

Statistical analysis was conducted using STATA v16.1 and XLStat v23.1.1

Results

A total of 555 patients underwent cardiac surgery in our institution during weeks 12–37 of 2020.

Of this original cohort, 326 (59%) underwent one of the three procedures (CABG, AVR or AVR&CABG). One patient from the COVID-19 era was excluded because the EuroSCORE II data were incomplete. In the corresponding epochs of the previous five years (the pre-COVID-19 era), 1692 patients underwent one of these three procedures in our institution. There were no missing data in the control group. Preoperative characteristics, pre-operative risk as predicted by EuroSCORE II, and procedures performed are displayed in Table 1.

Table 1.

Preoperative and operative characteristics of patients undergoing AVR, CABG or AVR&CABG during COVID-19 (2020) and pre-COVID-19 (2015–2019) eras.

	Pre-COVID-19	COVID-19
	(n = 1692)	(n = 325)
Age (years; median [IQR])	68 [61–75]	66 [59–72]
Gender (%)	male: 1326 (78)	male: 262 (80)
Hypertension (%)	1351 (80)	234 (72)
Diabetes (%)	559 (33)	126 (39)
Chronic airways disease (%)	229 (14)	20 (6)
Creatinine (µmol/L; median [IQR])	82 [70–96]	76 [66–92]
Body mass index (kg/m2, mean ± SD)	29 ± 4.9	29 ± 5.1
EuroSCORE II
Mean ± SD	2.37 ± 3.32	2.36 ± 4.70
Median [IQR]	1.39 [0.89–2.57]	1.38 [0.94–2.44]
Operative procedures	n (%)	n (%)
CABG alone	1176 (70)	237 (73)
AVR alone	367 (22)	66 (20)
AVR&CABG	149 (9)	22 (7)

AVR: aortic valve replacement; CABG: coronary artery bypass graft; IQR: interquartile range; SD: standard deviation.

Using inverse probability weighting, with average treatment effect weights, generated using the propensity scores, the outcomes in the COVID-19 era were compared with patients undergoing the same procedures in the equivalent epochs of each of the previous five years. The incidence of major complications (new renal failure necessitating dialysis/hemofiltration, new permanent or transient stroke and respiratory failure necessitating re-ventilation) in the two groups was compared using chi-square tests, weighted by the inverse probability weights for our sample. Table 2 shows the contingency tables for each major complication together with the p-value.

Table 2.

Incidence of major complications.

	Pre-COVID-19	COVID-19	p-value
	(n = 1692)	(n = 325)
New renal failure	46 (3%)	9 (3%)	0.923
No new renal failure reported a	1646	316
New stroke	10 (1%)	2 (1%)
No new stroke reported a	1682	323	0.991
New respiratory failure	43 (3%)	9 (3%)
No new respiratory failure reported a	1649	316	0.856

^aIncludes missing values.

For postoperative stay(days), 30-day and six-month overall survival, the proportional hazards assumption was tested, and verified to hold.

There was no difference in postoperative stay(days) between the two groups (sub-hazard ratio [SHR] = 1.00 [95% CI 0.90–1.12], competing-risks survival regression: p-value = 0.959). Figure 1 outlines the Kaplan-Meier curves for 30-day and six-month survival, including number at risk at each timepoint.

Figure 1.

Kaplan-Meier survival curves for COVID-19 era (2020; red line) and pre-COVID-19 (2015–2019; blue line).

Thirty-day survival was 98.3% (95% CI 97.6–98.9) in the pre-COVID-19 group and 98.8% (95% CI 96.8–99.5) in the COVID-19 era group (HR = 0.76 (95% CI 0.27–2.20), Cox-based test for equality: p-value = 0.6211). Six-month survival was 97.4% (95% CI 96.5–98.1) in the pre-COVID-19 group and 97.5% (95% CI 95.2–98.8) in the COVID-19 era group (HR = 0.94 [95% CI 0.44–2.01], Cox-based test for equality: p-value = 0.8809).

With respect to length of hospital stay, incidence of major complications, 30-day all-cause mortality and survival at six months there is no difference in the outcome of cardiac surgery in patients operated upon in the COVID-19 and pre-COVID-19 eras.

Discussion

This paper represents a snapshot of our experience of maintaining a cardiac surgical service through the first six months of the COVID-19 pandemic in the UK. The key findings of the study are that, for the index cardiac surgical procedures, the 30-day all-cause mortality, survival at six months, the incidence of major complications and lengths of hospital stay are the same as in the previous five years.

After a phase of justifiable caution and falling demand, we have developed a system for the safe conduct of cardiac surgery with specific strategies for the stratification of urgent referrals, waiting list management and infection control ¹ that have enabled our non-COVID work to continue at a time when the demand for some of our other specialist services has been unprecedented. ⁸

To the best of our knowledge, this study is the first to attempt to make an objective comparison between cardiac surgical outcomes in the COVID-19 and pre-COVID-19 eras. We have restricted our analysis to the so-called index procedures, generally accepted as surrogates for all adult cardiac surgery, when making comparison between individual surgeons and institutions.^9,10 No patient was lost to follow-up. In terms of limitations, the study is, of necessity, retrospective and has been restricted to relatively low-risk procedures; caution should be exercised in extrapolating these conclusions to higher-risk cardiac surgical procedures, or indeed other surgical disciplines.

In selecting data from five years ago as a control group for comparison with data from 2020, we could potentially have overlooked an epoch effect in which, for the sake of argument, the outcome of all cardiac surgery had improved significantly between 2015 and 2019, rendering invalid a comparison between 2020 and 2015. We considered this possibility when planning our study. The risk-adjusted hospital mortality for all cardiac surgery in the UK declined steadily between 2002 and 2016. ¹¹ However, the most recent NICOR report, which includes data up to and including 2018/19, shows that the crude unadjusted mortality has not changed materially between 2016 (2.41%) and 2019 (2.59%). ¹² We reasoned therefore that there is unlikely have been an ‘epoch effect’ between 2016 and 2020 that would materially impact our analysis whereas there would be a benefit in having more data with which to make comparison with the study group.

The World Health Organization (WHO) declared the COVID-19 outbreak to be a global pandemic in March 2020. ¹³ On 3 March, NHS England declared COVID-19 outbreak to be a ‘Level-4 National Incident’ permitting NHSE to take control of all healthcare resources, including those in the private sector. This designation has been lifted and re-imposed at different times since but, in common with healthcare systems the world over, much of the focus of the NHS has been on preparing for and managing the expected surges in COVID-19 admissions as the pandemic has waxed and waned in parallel with the imposition and easing of distancing measures to prevent its spread. Much has been written in both the lay and medical press of the challenges posed by attempting to continue to provide the expected standard of care for all patients at a time when a substantial proportion of healthcare resources has been diverted for the management of a global pandemic. It is difficult to find a comparable peacetime historical equivalent: the influenza pandemic of 1918 is often cited but the hospital medicine of 1918 is scarcely comparable with that of 2021. ¹⁴

As a specialist cardiothoracic hospital without an accident and emergency department, we have been faced with the challenge of maintaining both acute and elective cardiac surgical services at a time when much of the capacity of the hospital, indeed of the entire healthcare system, has been taken up with treating patients whose primary diagnosis was COVID-19. We observed a dramatic decline in the number of acute referrals at the start of the pandemic as well as at the start of subsequent surges, an observation which has been corroborated across the UK^15,16 and elsewhere. ¹⁷ This was undoubtedly a reflection of capacity constraints but when interviewing patients by telephone, we also observed an understandable reluctance on their part to be admitted for elective surgery.^18,19 We sought therefore to obtain objectively verifiable data which confirm that the outcomes of cardiac surgery in the COVID-19 era are comparable with those of previous eras and thereby reassure patients that it is safe for them to be admitted to hospital for elective surgery.

In the UK, the ONS publishes data on death registrations compared, on a weekly basis, with the average number of deaths for the corresponding epoch in the previous five years. While the increase in death rates due to COVID-19 has been extensively discussed in both the scientific and lay press, the excess mortality associated with non-COVID diagnoses has received less attention.

ONS data from 13 March–24 April 2020 (weeks 12–18) reveal an all-cause mortality in the UK that was approximately double the five-year average for the corresponding timeframe. ²⁰ The number of deaths occurring outside of hospitals exceeded the average by about 8000 and 80% of those deaths were unrelated to COVID-19, implying a substantial excess mortality in the community due to non-COVID-19 diagnoses. Figure 2 is derived from ONS data showing an excess all-cause mortality when compared with the same epoch of the previous five years, and an excess mortality for non-COVID diagnoses during weeks 13–18, 20 and 33–35 (Note: Reproduced from data downloaded from the website of the ONS). Of course, this increased mortality due to non-COVID diagnoses may be a reflection of capacity constraints but the finding lends weight to our impression of a reluctance on the part of patients to be admitted to hospital.

Figure 2.

Office for National Statistics registered deaths during weeks 12–37, 2020 (incl.).

At the start of the pandemic, these fears were not confined to patients and their families. Anecdotal reports from Italy suggested that contracting COVID-19 while hospitalised for cardiac surgery was associated with a very poor prognosis, something which seemed intuitively correct when considering the co-morbidities and risk factors known to be associated with a poor outcome in COVID-19 ²¹ ; some physicians were concerned that, irrespective of capacity constraints, admitting patients to hospital for all but the most urgent and emergency procedures would risk exposing them to undue hazard.

These concerns have been eloquently summarised by Rosenbaum ²² in an article in the New England Journal of Medicine: ‘Should hospitals schedule LVAD (left ventricular assist device) placements when ICU (intensive care unit) and ventilator capacity may soon be exceeded? Is a patient with severe aortic stenosis more likely to die from his underlying valvular disease or from a valve-replacement hospitalisation that leaves him with coronavirus infection?’

In our planning discussions at the start of the pandemic, at a time when all elective surgery had been suspended, we agreed that we should continue to attempt to offer patients the best guideline-directed treatment for their pathology as opposed, for example, to systematically opting for percutaneous options in the interests of expediency.

Conclusion

We believe this to be the first report of cardiac surgical outcomes in the COVID-19 era in which an attempt is made at objective comparison with previous eras. We conclude that for the most commonly performed index procedures (CABG, AVR, CABG&AVR) the outcome, based on the incidence of major acute complications, 30-day all-cause mortality and survival at six months, is the same in the COVID-era as in the previous five years.

We believe therefore that we can, at least, answer the second of the two questions posed by Rosenbaum’s ²² thought-provoking piece in the affirmative: patients can be safely consented for surgery on the basis that their risk of death and/or major complications is no higher during the COVID-19 pandemic than in the previous five years and should be discouraged from delaying seeking medical advice or deferring elective surgery because of understandable concerns about the transmissibility of COVID-19. All involved in healthcare delivery, patients and the wider public should be substantially reassured by these findings.

Declarations

Competing Interests

None declared.

Funding

None declared.

Ethics approval

The data forming the basis of the study were taken from a standard departmental audit and, as such, it was not a research study in the strict sense of the term.

Guarantor

JG.

Contributorship

JG, UAS, FF designed the study and directed the project. HS and MA were responsible for data collection. ED and FF were responsible for devising the statistical method and the statistical analysis. JG wrote the first draft of the manuscript. All authors contributed equally to the revision and editing of the final draft.

Provenance

Not commissioned; peer-reviewed by Abdulaziz Alkhulaifi and Julie Morris.

ORCID iD

Jullien Gaerhttp://orcid.org/0000-0002-6618-3730

Supplemental material:

Supplemental material for this article is available online.