Post-acute mortality following COVID-19 infection compared to matched test-negative people in England: findings from the ECHOES national surveillance system

Background: Mortality risk in the acute phase of SARS-CoV-2 infection has been well characterised, however longer-term risks associated with COVID-19 remain poorly understood.
Here, the Evaluation of Post-Acute COVID-19 Health Outcomes (ECHOES) dataset is used to investigate post-acute COVID-19 mortality in England. ECHOES is a nationwide cohort of individuals testing for SARS-CoV-2 designed to investigate post-acute COVID-19 health outcomes by linking test records to multiple real world data sources.

Objectives: To estimate risk of post-acute death >28 days to 1 year after SARS-CoV-2 infection, relative to those testing negative.

Methods: Data was obtained via linkage of COVID-19 testing data, the national vaccine registry, death registration data and Hospital Episode statistics data (HES). Positive and negative SARS-CoV-2 tests from May 2020 to April 2022 were matched 1:1 on week of test, sex, age band, index of multiple deprivation and area of residence. Follow up began at earliest test and was censored at end of study and death date, with negatives also censored if they subsequently tested positive. Individuals could contribute to both testing groups.
Using this the matched cohort, rate of death was calculated, and Cox proportional hazards models stratified by matched sets were used to assess risk after adjustment for ethnicity, acute severity and comorbidity indicators. Severity was determined by hospital attendances within -7 to +14 days of test, categorised as A&E attendance, hospital admission, or critical care, prioritising most severe category. Attendances for planned procedures, pregnancy, injury, or poisoning- were dismissed . Comorbidities diagnosed prior to test were categorised using the Charlson comorbidity index (CCI) as mild, moderate, or severe.

Results: The matched cohort comprised 8,419,574 positives and 8,419,574 negatives. Post-acute rate of death was 9.76 per 1,000 in positives and 5.92 in negatives, with an absolute rate difference (ARD) of 3.83. The unadjusted Cox model produced a Hazard Ratio (HR) of 1.77 (95% CI:1.75-1.80). After adjustment, the HR attenuated to 1.68 (95% CI: 1.65-1.71). The highest HRs were in females (HR:1.81, 95% CI:1.77-1.86; ARD 4.03), and those aged over 80 (HR: 2.07, 95% CI:2.02-2.12; ARD:118.89) with no difference seen in those under 50 (HR:0.98, 95%CI:0.90-1.07; ARD:0.10).

Conclusions: In this large, national cohort, we see a 70% increase in risk of post-acute death up to one year after testing positive for SARS-CoV-2, accounting for comorbidities and demographic and disease characteristics. This association is stronger in women and older people, with no increased risk in those aged under 50. These findings emphasise the importance of continued research into long-term morbidity and mortality of COVID-19.