This report is the first to present a comprehensive view of the similarities and differences between patients with COVID-19 managed with VV-ECMO in the 1st and 2nd waves of the pandemic from referral to follow-up. We have shown that patients with COVID-19 supported on VV-ECMO in the 2nd wave had an increased mortality and longer ECMO runs than patients managed in the 1st wave. We have confirmed the results of our previous work (4), which revealed that the patients who survive their ECMO run for COVID-19 also tend to survive in the long term. We have also, for the first time as far as we can tell, demonstrated that pulmonary function and quality of life is similar, between survivors of the 2 waves at an average of 7 months follow up.
In the UK, ICNARC reported a new increase in ICU admissions due to COVID-19 after 01/09/2020, defining the 2nd wave of the pandemic (16). The results of several studies, particularly the REMAP-CAP trial, have changed the standard of care for patients with COVID-19 admitted to hospital between the 1st and 2nd waves. Dexamethasone (8), and anti-IL-6 drugs (9, 21), have been shown to improve mortality, whilst remdesivir (22) reduced time to recovery. The adoption of these treatments is well represented in our baseline characteristics. These interventions have undoubtedly improved global outcomes in patients presenting to hospital with COVID-19 (16) but are not universally effective in preventing some patients from requiring extra-corporeal support.
Patients in the 2nd wave had lower inflammatory markers at cannulation, likely due to an alteration in host response to different COVID-19 variants and previous steroid treatment (23). Despite a dampening of the measured inflammatory response they continued to deteriorate to require VV-ECMO. They were also more likely to receive a pulse of methylprednisolone once on ECMO, for non-resolving COVID-19 pneumonitis. Whilst the optimum timing of immunomodulation remains uncertain (24) our data suggests that patients in the 2nd wave were less likely to respond to steroid treatment but were still susceptible to their side-effects (12). This may have had a significant impact on the duration of ECMO run and mortality seen in the 2nd wave.
A further difference seen between the 1st and 2nd waves is a reduction in the incidence of PE. This is probably due to more patients receiving treatment dose anticoagulation very early in the disease (25). This along with the increase use of immunomodulators (8, 9) may explain the reduction in documented PEs. This reduction in burden of vascular disease in the 2nd wave may also be a factor in the significant difference in KCOc between the 1st and 2nd waves at follow up.
As well as changes in pharmacological interventions, studies have suggested that NIV could prevent intubation and reduce mortality in COVID-19 ARDS (10). We found a longer duration of NIV in patients supported on ECMO in the 2nd wave. These findings were mirrored across the rest of the National ECMO service (17). Duration of invasive mechanical ventilation prior to ECMO represents a well-established risk factor for increased mortality (26), whilst duration of NIV is less well studied. Our findings suggest that in COVID-19 acute respiratory distress syndrome (ARDS) duration of NIV but not invasive mechanical ventilation may be associated with reduced survival in patients going on to require VV-ECMO. These findings are supported in the wider literature (27, 28).The increased duration of NIV in patients in the 2nd wave meant these individuals may have been more likely to be exposed to self-induced lung injury and delayed institution of prone positioning (29). This hypothesis is supported by the association between barotrauma and NIV duration in our cohort. Our work highlights the need for urgent studies of the optimal timing of intubation in COVID-19 ARDS.
The survival data, although showing a difference in outcome between the 1st and 2nd waves to date of censure and decannulation also suggest that very few patients who survived to 90-days died during the follow up period. This finding was independent of whether the patient was supported in the 1st or 2nd wave and is in keeping with data on longer term survival after ECMO from other causes of ARDS (1). The increased mortality was despite a decrease in SOFA and Murray scores seen in the 2nd wave. Importantly, traditional ICU severity scores have been shown to be poorly predictive of outcomes in COVID-19 related critical illness (30).
Despite the reduction in survival between the 1st and 2nd waves, mortality in both groups remains similar to historical cohorts (3, 4). The better survival rates reported in our cohort than in the general literature could be partly due to the high volume of patients supported in our center, a factor associated with low mortality in other studies (31). In contrast to our data a study reporting the outcomes of patients referred but deemed not suitable for ECMO has shown a mortality of 80% at 90 days (32). With mortality in our 2nd wave cohort of just more than 30%, our findings highlight the importance of VV-ECMO in supporting a carefully selected group of patients with COVID-19 ARDS.
As well as showing the differences in outcome over the pandemic our data also sheds some light on who might benefit most from support with VV-ECMO for COVID-19. Age is an important predictor of mortality in patients with COVID-19 who are managed with VV-ECMO both in our study and the wider literature (3). Whilst there is no upper age-limit to consider ECMO in COVID-19 it remains an important consideration, with the original studies of ECMO excluding patients over 65 years of age (1). The RESP score was developed to predict outcomes on ECMO (26) and despite the narrow score range seen in our population, was still significantly higher in those who survived compared to those who died. Our data therefore validates the RESP score as a tool which can predict survival on ECMO even within a narrow window.
The median duration of ECMO was 11 days longer in the 2nd wave than the 1st. This has been verified in a number of other studies (33). As well as implication for individuals and families these longer runs have economic implications on healthcare provision with every day of ECMO support costing over $7500 in one study (34), making it a huge financial burden particularly in resource poor healthcare settings.
Follow up data on patients supported on VV-ECMO for COVID-19 is lacking (14). Here we present a well characterized unique cohort of patients who had lung function and quality of life measured at an average of between 7 and 8 months after the initiation of their ECMO run. We have demonstrated that there is very little difference in the follow up parameters of patients supported in the 1st and 2nd waves of the pandemic. This suggests that despite the reduction in survival across the 2 waves patients who survive to follow up have similar physiological reserve and quality of life. This adds weight to the argument that ECMO is a worthwhile treatment in a carefully selected group of patients with COVID-19. Whilst there was no statistical difference in QOL between the 2 waves there were some numerical differences in the proportion of patients reporting mobility, self-care and usual activity issues. This may relate to an increase in critical care-neuro-myopathy precipitated by a longer duration of ECMO run, ICU stay and increased use of steroids (35). The QOL in our survivors is typical of that seen in patients after ARDS due to COVID-19 and other causes (36, 37) and is lower, numerically at least, than a healthy population of the same age who have an expected EQ5D3L index score of 0.8 (38). As well as the aforementioned increased KCOc in the 2nd wave the lung function parameters at follow up in our group as a whole show a disproportionate reduction in transfer factor of carbon monoxide (TLCOc) with relatively preserved spirometry. This is in keeping with follow up data from the wider ARDS and COVID-19 literature (36, 37).