Mortality following recovery from COVID-19 hospitalization: a long-term cohort study

doi:10.21203/rs.3.rs-4999598/v1

Download PDF

Article

Mortality following recovery from COVID-19 hospitalization: a long-term cohort study

https://doi.org/10.21203/rs.3.rs-4999598/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

The COVID-19 pandemic has profoundly affected public health worldwide. Although research shows that COVID-19's health effects can persist for up to two years, the long-term impact on mortality in patients who survived the acute phase remains unclear. We investigated the association between COVID-19 hospitalization and all-cause mortality up to 3.5 years in these patients.

Methods

The study included Clalit Health Services members aged 40 years or older. The case group comprised members hospitalized due to COVID-19 between March 2020 and December 2021. Controls who did not contract COVID-19 during this period were matched at a 1:1 ratio based on birth year, sex, and Charlson comorbidity score. Follow-up began 30 days after discharge from the last COVID-19 hospitalization and ended on 30 September 2023, or upon all-cause mortality, whichever occurred first. Data were analyzed using Kaplan-Meier curves and multivariable frailty Cox regression models. Stratification was performed by age groups 40–64 years and 65 years or older.

Results

The incidence of all-cause mortality was higher compared to controls (4.91 vs. 2.63 per 1,000 participants, respectively), with a hazard ratio (HR) of 1.69 (95% CI, 1.57–1.83). Among patients aged 40–64, the HR for mortality was higher than in those aged 65+, at 2.31 (1.79–2.98) compared to 1.63 (1.50–1.76), respectively.

Conclusions

The study demonstrates that patients who have survived COVID-19 hospitalization have an elevated risk of all-cause mortality up to 3.5 years. In light of these findings, it is crucial to consider initiating follow-up for individuals hospitalized due to COVID-19, particularly those in the 40–64 age group, to address their ongoing health needs adequately.

Health sciences/Diseases/Infectious diseases/Viral infection

Health sciences/Health care/Health policy

Health sciences/Medical research/Epidemiology

Health sciences/Risk factors

Long COVID, or post-acute sequelae of SARS-CoV-2 infection (PASC), is a chronic illness that can develop following SARS-CoV-2 infection, lasting at least 3 months with continuous, relapsing, remitting, or progressive symptoms affecting one or more organ systems.¹ The severity of the initial infection, particularly in hospitalized patients, is a strong predictor of long-term health consequences, including mortality.²

The COVID-19 pandemic has caused more than 775.8 million confirmed cases across the world and 7.1 million deaths as of 24 July 2024.³ Several observational studies have examined the association between COVID-19 and all-cause mortality in the short-term^4,5 and in the long-term,⁶ but fewer studies have assessed this association in hospitalized patients during the acute phase of the disease in the long-term.⁷ Most studies assessing the long-term association have been based on data from the U.S. Department of Veterans Affairs healthcare system (VA), yielding mixed results.^7–9

One study reported an elevated risk of death in the second year (relative risk (RR) of 1.29)) among COVID-19 patients hospitalized during the acute phase, with null association in non-hospitalized patients (RR = 0.99).⁹ However, another study of VA members found a decreased mortality risk (HR = 0.89) during the second year post-infection.⁸ The most recent study using the same VA data reported the same RR of 1.29 after three years of follow-up in hospitalized patients.⁷ These studies of the VA members enrolled older (mean age > 63 years), predominantly White (> 70%) males (> 90%), and only included individuals who tested positive for SARS-CoV-2 before the introduction of vaccines. Therefore, our study aimed to evaluate the risk for all-cause mortality up to 3.5-year post-hospitalization in a diverse population, including vaccinated individuals.

Study population

We identified 16,445 Clalit members aged 40 years or older who were alive 30 days after a COVID-19-related hospitalization from 1 March, 2020, to 31 December, 2021. All patients were matched 1:1 with 16,445 control Clalit members without positive SARS-CoV-2 test results during that period, using exact matching by birth year, sex, and Charlson comorbidity score, resulting in a total of 32,890 study participants (Fig. 1).

Data for all variables was fully available for the entire study population. COVID-19 hospitalized patients had shorter follow-up duration (28.3 vs. 29.6 months) compared to controls. Patients had higher rates of comorbidities compared to controls, including cancer (20.0% vs. 15.6%), diabetes (41.3% vs. 34.1%), HF (11.5% vs. 6.4%), renal disease (14.3% vs. 9.6%), COPD (8.7% vs. 6.6%), and dementia (8.6% vs. 5.5%). They were more likely to be smokers (21.6% vs. 7.8%) and had a higher BMI (mean: 29.8 vs. 28.2 kg/m²). Additionally, the COVID-19 group had a lower SES score (mean: 4.8 vs. 5.7) (Table 1).

Table 1

Characteristics of the study participants (N = 32,890)
Characteristic	Overall, N = 32,890¹	Cases, N = 16,445¹	Controls, N = 16,445¹
Death during follow-up period	3,702 (11.3%)	2,373 (14.4%)	1,329 (8.1%)
Follow-up period	29.0 (8.2)	28.3 (8.9)	29.6 (7.2)
Sex, Female	15,986 (48.6%)	7,993 (48.6%)	7,993 (48.6%)
Charlson score	2.6 (2.7)	2.7 (2.7)	2.6 (2.7)
Age, years	66.0 (14.2)	66.0 (14.2)	66.0 (14.2)
SES	5.2 (2.1)	4.8 (2.0)	5.7 (2.1)
Population group
General population	22,484 (68.4%)	9,898 (60.2%)	12,586 (76.6%)
Arabs	8,706 (26.5%)	5,378 (32.7%)	3,328 (20.2%)
Ultra-orthodox	1,684 (5.1%)	1,163 (7.1%)	521 (3.2%)
Smoking status
Non-smokers	21,738 (66.7%)	11,644 (71.6%)	10,094 (61.8%)
Past smokers	6,883 (21.1%)	3,355 (20.6%)	3,528 (21.6%)
Smokers	3,990 (12.2%)	1,268 (7.8%)	2,722 (16.7%)
BMI	29.0 (5.9)	29.8 (6.2)	28.2 (5.5)
Cancer	5,854 (17.8%)	2,562 (15.6%)	3,292 (20.0%)
Diabetes	12,405 (37.7%)	6,794 (41.3%)	5,611 (34.1%)
IHD	7,273 (22.1%)	3,874 (23.6%)	3,399 (20.7%)
CHF	2,935 (8.9%)	1,886 (11.5%)	1,049 (6.4%)
Renal failure	3,930 (11.9%)	2,357 (14.3%)	1,573 (9.6%)
COPD	2,506 (7.6%)	1,425 (8.7%)	1,081 (6.6%)
Asthma	2,649 (8.1%)	1,355 (8.2%)	1,294 (7.9%)
Depression	4,502 (13.7%)	2,340 (14.2%)	2,162 (13.1%)
Dementia	2,314 (7.0%)	1,413 (8.6%)	901 (5.5%)
Number of COVID-19 immunization shots
0	20,691 (62.9%)	12,565 (76.4%)	8,126 (49.4%)
1	2,303 (7.0%)	1,470 (8.9%)	833 (5.1%)
2	6,741 (20.5%)	1,947 (11.8%)	4,794 (29.2%)
3	3,155 (9.6%)	463 (2.8%)	2,692 (16.4%)
Number of Hospitalizations in 3 years prior Covid-19 pandemic	1.1 (2.2)	1.3 (2.5)	0.8 (1.7)

Association between COVID-19 hospitalization recovery and long-term mortality

All-cause mortality occurred in 2,373 out of 16,445 cases (4.91 per 1000 person-days), compared to 1,329 out of 16,445 matched controls (2.63 per 1000 person-days) (Fig. 1). The multivariable model revealed a higher event rate in COVID-19 patients (Hazard Ratio (HR): 1.69; 95% Confidence Interval (CI), 1.57–1.83) (Table 2).

Table 2

Associations between participant characteristics and all-cause mortality following COVID-19 hospitalization
Variables	Results of the Univariable models HR (95% CI)	Results of the Multivariable model HR (95% CI)
Positive Covid-19	1.98 (1.85, 2.12)	1.69 (1.57–1.83)
Age	1.08 (1.08, 1.08)	1.07 (1.06–1.07)
Gender	1.02 (0.87–1.20)
SES	0.96 (0.94–0.97)	0.97 (0.96–0.99)
Population group (Ref = total population)
Arabs	0.97 (0.89, 1.06)
Ultra-orthodox	0.95 (0.81–1.11)
Smoking status (Ref = never)
Smoked in the past	1.03 (0.95–1.12)	1.05 (0.96–1.14)
Now smoking	1.38 (1.24–1.25)	1.56 (1.39–1.74)
BMI	0.98 (0.97–0.98)	0.97 (0.97–0.98)
Oncology	1.15 (1.07–1.24)	1.65 (1.54–1.78)
Diabetes	1.13 (1.05–1.22)	1.27 (1.18–1.37)
IHD	1.30 (1.21–1.39)	1.07 (1.00–1.16)
CHF	2.08 (1.92–2.25)	1.72 (1.58–1.87)
Renal failure	1.76 (1.62–1.91)	1.66 (1.54–1.80)
COPD	1.48 (1.35–1.62)	1.21 (1.10–1.33)
Asthma	0.96 (0.86–1.07)
Depression	1.28 (1.19–1.39)	1.15 (1.05–1.24)
Dementia	2.42 (2.23–2.63)	2.10 (1.93–2.29)
Hospitalizations in 3 years prior Covid-19 pandemic	1.11 (1.10–1.12)	1.07 (1.06–1.08)
Number of COVID-19 immunization shots (≥ 2)	0.84 (0.77–0.90)	0.89 (0.79–1.01)
Months since first vaccine dose	1.01 (0.99–1.02)	1.03 (1.01–1.05)
Footnotes:

Several comorbidities were significantly associated with an increased risk of all-cause mortality following COVID-19 hospitalization. Cancer (1.65 (1.54–1.78)), diabetes (1.27 (1.18–1.37)), CHF (1.72 (1.58–1.87)), renal failure (1.66 (1.54–1.80)), COPD (1.21 (1.10–1.33)), depression (1.15 (1.05–1.24)), and dementia (2.10 (1.93–2.29)) were all associated with higher mortality risk.

In contrast, the number of COVID-19 immunization doses (≥ 2) was associated with a lower, though not statistically significant, risk of mortality (0.89 (0.79–1.01)).

Sub-group Analysis

The risk for all-cause mortality differed by age group, with cases aged 40–64 having a higher HR of 2.31 (1.79–2.98) compared to 1.63 (1.50–1.76) in cases aged 65 years or older (Table 3). Smoking, lower BMI, and presence of chronic conditions (cancer, diabetes, CHF, renal failure, COPD, depression, and dementia) were associated with increased mortality risk in both age groups. The association between cancer and renal failure and the risk for early mortality was more pronounced in the younger cohort.

Table 3

Multivariable Cox-regression model for assessing the association between COVID-19 hospitalization and all-cause mortality by age group
Variables	Ages 40–64 HR (95% CI)	Ages 65+ HR (95% CI)
Positive Covid-19	2.31 (1.79–2.98)	1.63 (1.50–1.76)
Age	1.05 (1.03–1.07)	1.07 (1.06–1.07)
SES	0.95 (0.90–1.01)	0.98 (0.96–1.00)
Smoking status (Ref = never)
Smoked in the past	0.95 (0.72–1.26)	1.06 (0.97–1.16)
Now smoking	1.38 (1.03–1.84)	1.58 (1.40–1.79)
BMI	0.98 (0.96–0.99)	0.97 (0.97–0.98)
Oncology	4.61 (3.69–5.77)	1.49 (1.38–1.61)
Diabetes	1.85 (1.46–2.33)	1.22 (1.13–1.32)
IHD	0.82 (0.61–1.11)	1.09 (1.01–1.18)
CHF	2.45 (1.75–3.43)	1.67 (1.53–1.83)
Renal failure	2.18 (1.63–2.90)	1.60 (1.48–1.74)
COPD	1.52 (1.09–2.11)	1.20 (1.08–1.33)
Depression	1.40 (1.06–1.85)	1.12 (1.03–1.22)
Dementia	1.93 (1.06–3.50)	2.10 (1.92–2.28)
Hospitalizations in 3 years prior COVID-19 pandemic	1.11 (1.08–1.13)	1.06 (1.05–1.07)
Number of COVID-19 immunization shots (≥ 2)	0.52 (0.32–0.84)	0.92 (0.81–1.05)
Months since first vaccine dose	1.12 (1.05–1.20)	1.03 (1.01–1.05)

For ages 40–64, receiving two or more COVID-19 immunization doses was associated with a significantly reduced risk of mortality (0.52 (0.32–0.84)). In ages 65 and older, the association was weaker and not statistically significant (0.92 (0.81–1.05)).

The risk of mortality was higher in males than in females (1.75 (1.57–1.96) vs. 1.55 (1.39–1.73)) (Table S1).

Landmark Analysis

The excess risk for all-cause mortality in cases compared to controls decreased over time from 1.53 (1.40–1.67) at time point of 6 months reaching 1.28 (0.66–2.47) by the end of the follow-up period, as demonstrated by the landmark analysis results (Figure S1). Despite the decrease in excess risk, the mortality risk associated with COVID-19 infection remained significantly elevated throughout follow-up.

In this study of 16,445 hospitalized patients due to COVID-19 illness and 16,445 controls followed for up to 3.5 years after hospitalization discharge, we found that the risk for all-cause mortality was higher in hospitalized patients compared to matched controls (1.69 (1.57–1.83)). This could be crucial for developing targeted interventions or understanding the long-term impact of the COVID-19 outbreak, and underscores the importance of long-term follow-up and care for individuals who experienced severe COVID-19.² Although this risk decreased over time, it remained persistent even after 3.5 years. The risk was higher in patients aged 40–64 (2.31 (1.79–2.98)) compared to those aged 65 and older (1.63 (1.50–1.76)). Receiving two or more COVID-19 vaccination doses was associated with a reduced risk in the 40–64 age group (0.52 (0.32–0.84)) compared to those aged 65 and older (0.92 [0.81–1.05]).

Our findings contrast with those of Iwashyna et al., who reported a protective effect on mortality among COVID-19 survivors between days 181 to 365 (HR, 0.92 (0.89–0.95)) and days 366 to 730 (0.89 (0.85–0.92)).⁸ The risk in our total population is higher than that reported in Cai et al., a recent study, which followed a large cohort of hospitalized COVID-19 patients and matched controls for 3 years (Incidence rate ratio: 1.29 (1.19–1.40))⁷. Those difference may be due to the characteristics of the study participants. The aforementioned studies were based on VA members, who are mostly aged over 65, predominantly male, and mostly White, and do not fully represent the general population's diversity.

The present study has several strengths that enhance its validity and impact. Firstly, our cohort is highly generalizable to the broader population, being representative of the various population groups in Israel, including sex, age groups, population groups, and socioeconomic status. This diversity allows us to control for potential confounders more effectively, ensuring that our findings are applicable across different demographic segments. Moreover, we considered the number of vaccine shots administered to each participant and controlled for it in our analysis. Vaccine status, strongly associated with SARS-CoV-2 infection and mortality, was a crucial factor in our study. By controlling for vaccination, we were able to yield more accurate estimates of the long-term effects of COVID-19.

Additionally, our study features a relatively long follow-up period extending up to 3.5 years, which, to the best of our knowledge, is the most extended period studied to date in the context of the long-term effects of COVID-19. This extensive follow-up enables us to provide more comprehensive and informative insights into the enduring impact of this novel virus on long-term health outcomes. Furthermore, the large size of the Clalit member database allowed us to select a control group through exact matching, thereby reducing the potential for misclassification bias. Using a frailty Cox regression model further enhanced our analysis by accounting for the inherent variability within clusters.

This study also has several limitations. Participants who were hospitalized due to COVID-19 during the Omicron wave were censored at the date of hospitalization in our study. However, those who tested positive for SARS-CoV-2 during the Omicron period but were not hospitalized could still experience mortality later on. If these subsequent deaths are erroneously attributed to a previous infection rather than the Omicron variant, it could lead to misclassification of the cause of death. This misattribution may distort the analysis by overestimating the mortality risk associated with earlier infections and underestimating the impact of the Omicron variant, which is known to have a different clinical course compared to earlier strains. Properly accounting for the variant and timing of infection is essential to ensure accurate conclusions about the long-term mortality risks associated with COVID-19.

Additionally, during the study period, older individuals and those with chronic diseases were prioritized for vaccination, potentially introducing bias when estimating mortality risk following COVID-19. This prioritization could lead to a differential impact on mortality rates, with vaccinated participants, particularly those at higher risk, experiencing a reduced mortality risk. This temporal difference could introduce bias into the study, as the protective effects of the vaccines might reduce the mortality risk for later participants, skewing the comparison between early and later entrants.

The findings of this study underscore the significant long-term impact of COVID-19 hospitalization on all-cause mortality, highlighting the importance of sustained healthcare support for survivors even after discharge. Policymakers should consider implementing comprehensive post-discharge monitoring and care programs to mitigate the elevated mortality risk observed in this population. This could involve regular follow-ups, particularly for vulnerable groups such as younger ages and those with pre-existing conditions. Additionally, these findings reinforce the urgency of preventive measures, including vaccination and public health interventions, to reduce the severe COVID-19 requiring hospitalization.

In conclusion, this study provides compelling evidence that individuals hospitalized due to COVID-19 face an increased risk of all-cause mortality that extends well beyond the acute phase of the disease. Our findings contribute to a deeper understanding of the pandemic’s long-term impact on public health and underscore the critical need for ongoing research, targeted interventions, and policy measures to support the recovery and well-being of COVID-19 survivors.

Ethics statement

The study was approved by Clalit's Community Helsinki and Data Utilization committees. The study was exempt from the requirement to obtain informed consent owing to the retrospective design.

Study Design and Period

An observational, retrospective cohort study was conducted.

The study exposure period commenced on 1 March, 2020, coinciding with the onset of the COVID-19 outbreak in Israel, and concluded on 31 December, 2021 (prior to the Omicron variant wave). Throughout the study period, the primary circulating SARS-Cov-2 variants were the original Wuhan strain and the Alpha and Delta variants and their sub-variants.

The study outcome was all-cause mortality during the follow-up period, which ended on 30 September, 2023.

Data sources

The study was based on data obtained from the electronic medical records of Clalit, the largest of four integrated healthcare organizations in Israel, which insures 4.7 million patients (52% of the population).

Study Population

All Clalit members aged 40 years or older as of 1 March, 2020 were eligible for the study (n = 1,750,361) (Fig. 1). Patients who were hospitalized due to COVID-19 between March 2020 and December 2021 (16,445) were exactly matched, without replacement, to uninfected controls based on their year of birth, sex, and Charlson comorbidity score.

The individual index date was defined as 30-days after the last COVID-19-related hospitalization discharge, occurring between March 2020 and December 2021.

Follow-up ended on 30 September, 2023, upon all-cause mortality, or in the case of a COVID-19 hospitalization during the Omicron wave, whichever occurred first. For uninfected controls, follow-up started on the index date of the matched-case. The study population selection flow diagram is available as Fig. 1.

Covariates

A set of pre-defined covariates was selected based on prior knowledge of potential confounders that could influence the association between SARS-CoV-2 infection and mortality. Socio-demographic covariates included age, sex, population group (categorized as Arabs, Ultra-orthodox, and general population), socio-economic status (SES) (Appendix), and smoking status (classified as never, former, or current smokers). Comorbidities considered in the analysis included the last body mass index (BMI) measurement before the index date, diabetes, ischemic heart disease (IHD), congestive heart failure (CHF), renal disease, chronic obstructive pulmonary disease (COPD), asthma, depression, dementia, cancer. All the Comorbidities were sourced from the Clalit's chronic illness registry. The registry thoroughly reviews and scans the insured individual's entire medical history. Healthcare utilization variables included the number of hospitalizations in the 3 years prior to COVID-19 outbreak and the number of vaccine doses received until the index date.

Statistical Analysis

Descriptive statistics are given as means and standard deviations or as frequency (n) with percentage (%), according to the scale of the variable. We assessed the association between COVID-19- related hospitalization and all-cause mortality using the frailty-Cox proportional hazards model, with frailties common to pairs of case and control participants within the same cluster, defined by gender, birth year, and Charlson comorbidity index.¹⁰

Cox proportional hazards model assumptions were assessed using the log-log and Schoenfeld residual plots and the Schoenfeld's global test. When the assumptions of the model were violated, we used a Landmark analysis at 6-month intervals, as sub analysis post-hoc, to assess the time-varying impact of Covid-19 infection on all-cause mortality. Participants who survived and remained event-free at each 6-month interval were included in the subsequent Landmark analysis. Cox proportional hazards models were applied at each time point to estimate the hazard ratios for mortality, adjusting for age, comorbidities, and other relevant covariates.

A subgroup analysis by age (40–64 years, and 65 years or older) was pre-defined, and further validated as the interaction term between age and study group was significant in our final multivariable model.

R statistical software version 3.5.0 (R Foundation) was used for the univariate and multivariate survival analysis.

Clinical Overview of Long COVID. CDC. https://www.cdc.gov/covid/hcp/clinical-overview/index.html.
Al-Aly, Z. et al. Long COVID science, research and policy. Nature Medicine 2024 30:8 30, 2148–2164 (2024).
COVID-19 cases. WHO COVID-19 dashboard. https://data.who.int/dashboards/covid19/cases?n=c.
Liu, T. H. et al. Comparison of post-acute sequelae following hospitalization for COVID-19 and influenza. BMC Med 21, 1–10 (2023).
Oseran, A. S. et al. Long term risk of death and readmission after hospital admission with covid-19 among older adults: retrospective cohort study. BMJ (2023).
Msemburi, W. et al. The WHO estimates of excess mortality associated with the COVID-19 pandemic. Nature 2022 613:7942 613, 130–137 (2022).
Cai, M., Xie, Y., Topol, E. J. & Al-Aly, Z. Three-year outcomes of post-acute sequelae of COVID-19. Nature Medicine 2024 30:6 30, 1564–1573 (2024).
Iwashyna, T. J. et al. Late Mortality After COVID-19 Infection Among US Veterans vs Risk-Matched Comparators: A 2-Year Cohort Analysis. JAMA Intern Med 183, 1111–1119 (2023).
Bowe, B., Xie, Y. & Al-Aly, Z. Postacute sequelae of COVID-19 at 2 years. Nature Medicine 2023 29:9 29, 2347–2357 (2023).
Elghafghuf, A., Dufour, S., Reyher, K., Dohoo, I. & Stryhn, H. Survival analysis of clinical mastitis data using a nested frailty Cox model fit as a mixed-effects Poisson model. Prev Vet Med 117, 456–468 (2014).

There is NO Competing Interest.

Appendix.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Mortality following recovery from COVID-19 hospitalization: a long-term cohort study

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Introduction

Results

Study population

Association between COVID-19 hospitalization recovery and long-term mortality

Sub-group Analysis

Landmark Analysis

Discussion

Methods

Ethics statement

Study Design and Period

Data sources

Study Population

Covariates

Statistical Analysis

References

Additional Declarations

Supplementary Files

Status:

Version 1