Temporal Trends in ICU Survival for Coronavirus Disease 2019: A Bumpy Road.

Following the spring 2020 surge of coronavirus disease 2019 (COVID-19), many countries and hospital networks reported improved hospital and ICU survival by the summer months (1). For example, Dennis et al (2) reported that 30-day hospital survival in ICUs and high-dependency care units (HDUs) in United Kingdom increased by 26% and 21%, respectively, from March to June 2020. Later, Docherty et al (3) showed that overall hospital survival in the United Kingdom had increased from 68% to 84% during this same time frame. Furthermore, through mediation analysis, they were able to estimate that 10% of the improvement was due to changes in case-mix (i.e., lower disease severity and comorbidity burden) and 22% was due to changes in treatment (i.e., increased use of steroids and noninvasive ventilation)—whereas the majority was attributed to other unmeasured factors not included in the model (3).

In winter 2020–2021, the United Kingdom experienced wave 2 of COVID-19, fueled by the highly transmissible alpha (4) variant (B.1.1.7) of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). This new variant, coupled with the renewed strain on the U.K.’s National Health System, raised concerns that the previously observed improvements in hospital survival may not be durable.

In this issue of Critical Care Medicine, Dennis et al (5) build on their prior analysis (2) to examine survival from HDU and ICU COVID hospitalizations through wave 2. Nearly 50,000 adult patients admitted to an HDU or ICU in United Kingdom for COVID-19 (March 1, 2020, through January 31, 2021) were included in the study. The first major finding of their study was that—after a steady improvement in 30-day hospital survival during and after wave 1—hospital survival declined during wave 2 but never to the levels seen at the beginning of the pandemic. Temporal trends in mortality were similar across patient subgroups defined by age and comorbidity burden.

The pandemic has consistently highlighted the importance of critical care strain. However, given the temporospatial variation in COVID surges, daily hospital-level data are needed to measure the impact of strain on patient outcomes. For example, using granular daily ICU bed census data, Bravata et al (6) were able to show that the risk of hospital mortality among ICU patients was elevated (e.g., adjusted hazard ratio 1.67; 95% CI, 1.08–2.60; p = 0.05 1) when COVID ICU load was > 75% to 100%) in U.S. Veterans Affairs hospitals during March to November 2020. Fortunately, however, such severe strain was uncommon during the study period, particularly after April 2020.

Dennis et al (5) were able to leverage similarly granular data to adjust for critical care strain in their study in order to test whether the increased mortality during wave 2 was explained by critical care strain. For this analysis, strain was operationalized for each hospitalization as the percent of critical care bed occupancy at the treating hospital on the date of admission. (E.g., a hospital with 20 ICU-level patients but only 10 ICU beds prior to the pandemic would be classified as having 200% occupancy on that date.)

The second major finding of the study was that the increased risk of mortality during wave 2 in United Kingdom was only modestly attenuated after adjusting for critical care strain. For example, the adjusted hazard ratio for mortality in wave 2 among ICU-treated patients decreased from 1.88 (95% CI, 1.6–2.18) to 1.71 (95% CI, 1.45–2.01) after accounting for critical care strain and similar decreases in the strength of association were seen among HDU-treated patients (Table 1 in [5]). Because strain was measured slightly different from prior studies, and the magnitude of strain experienced was not reported, direct comparison with prior studies on the impact of strain is not possible (6). Nevertheless, their approach to measuring strain has strong face validity, and the persistent association of wave 2 with increased mortality even after adjustment for patient demographics, comorbidity burden, and strain suggests that additional factors must have contributed to the mortality difference.

The findings of the study by Dennis et al (5) are consistent with the possibility that the alpha (B.1.1.7) variant of SARS-CoV-2 is more deadly than other strains. Furthermore, in a study of over 1 million individuals diagnosed with SARS-CoV-2 in United Kingdom (November 1, 2020 through February 14, 2021) and for whom variant status was known, adjusted hazard ratio for mortality was 1.61 (95% CI, 1.42–1.82) for the alpha (B.1.1.7) variant compared with other strains—providing more direct evidence that this variant may be more deadly (7).

The temporal trends in mortality described in the study by Dennis et al (5) resonate beyond the United Kingdom. As the pandemic has evolved, it has become clear that long-term abatement from COVID-19 is not guaranteed. In our respective states, we too celebrated improved outcomes and waning case counts following wave 1, only to experience the situation deteriorate rapidly during subsequent waves. As with the current report, the reason for the tip in mortality is likely multifactorial. Until global vaccination reaches higher levels, we remain at risk from further variants which may drive mortality not only directly (by causing severe illness) but also indirectly (by causing excessive strain to the healthcare system).