ABSTRACT
INTRODUCTION: COVID-19-related organ dysfunction is increasingly recognized as sepsis, & sepsis has been reported as the most common proximate cause of death among COVD-19 patients in autopsy studies. Thus, the COVID-19 pandemic is expected to affect substantially the epidemiology of sepsis. However, the contribution of COVID-19 to sepsis-related mortality in the United States (US) is unknown. METHOD(S): We used the CDC WONDER Multiple Cause of Death database to identify decedents with a diagnosis of sepsis during 2015-2019 and with diagnoses of COVID-19, sepsis, or both during 2020. Sepsis was identified using previously reported ICD-10 code-based taxonomy. COVID-19 was identified by ICD-10 code U071. Negative binomial regression was used on the 2015-2019 data to forecast the number of sepsis-related deaths in 2020. We then compared the number of observed vs expected sepsis-related deaths in 2020. In addition, we examined the reporting of a diagnosis of COVID-19 among decedents with sepsis and the proportion of a diagnosis of sepsis among those with COVID-19. The latter analyses were then repeated across the Department of Health and Human Services (HHS) Regions. RESULT(S): In 2020, there were 242,630 sepsis-related deaths, 384,536 COVID-19-related deaths, & 35,057 deaths with both diagnoses. The expected number of sepsis-related deaths for 2020 was 207,175 (95% CI 205,929-208,429), with the ratio of observed to expected deaths 1.17 (95%CI 1.16-1.18). COVID-19-related deaths comprised 15.0% of all observed sepsis-related deaths, ranging from 8.1% (HHS Region 10) to 18.2% (HHS Region 2). A diagnosis of sepsis was reported in 9.1% of all COVID-19-related deaths, varying from 6.6% (HHS Region 2) to 12.5% (HHS Region 9). CONCLUSION(S): Sepsis-related mortality was reported in less than 1 in 10 COVID-19-related deaths in the US during 2020, with the frequency of sepsis diagnoses varying nearly 2-fold across HHS regions. Although the number of COVID-19-related deaths far exceeded sepsis-related mortality, the contribution of the former to the latter, based on death certificates, was relatively minor. Our findings suggest substantial underdocumentation and possibly underrecognition of sepsis among COVID-19 decedents, likely contributing to varying coding practices during the first year of the pandemic.
ABSTRACT
INTRODUCTION: COVID-19-related organ dysfunction is increasingly recognized as sepsis of viral origin and is a common complication among those requiring hospitalization, with estimated prevalence of over 50% among the latter. However, the population-level association of COVID-19 with short-term mortality among septic patients is unknown. METHOD(S): We used a statewide dataset to identify hospitalizations aged >=18 years with sepsis in Texas during April 1-December 31, 2020. Sepsis was defined by "explicit" and ICD-10 codes for severe sepsis (R65.20) and septic shock (R65.21) and COVID-19 by ICD-10 code U07.1. A hierarchical, mixed-effects model was fit to estimate the association of COVID-19 with short-term mortality (defined as in-hospital death or discharge to hospice) among sepsis hospitalizations. Sensitivity analyses of the sepsis hospitalization subsets with septic shock and ICU admission were performed using a similar modeling approach. RESULT(S): Among 55,145 sepsis hospitalizations, 13,149 (23.8%) had COVID-19. Compared to those without COVID-19, sepsis hospitalizations with COVID-19 were younger (aged >=65 years 53.6% vs 55.0%), more commonly male (59.5% vs 50.4%) and racial/ethnic minority (66.1% vs. 46.2%), with lower burden of chronic illness (mean [SD] Charlson comorbidity index 1.8 [1.9] vs 2.8 [2.6]), but with higher mean [SD] number of organ dysfunctions (3.1 [1.4] vs 2.7 [1.6]) [p < 0.0001 for all comparisons]. Short-term mortality among sepsis hospitalizations with and without COVID-19 was 52.7% vs 30.2%, respectively. On adjusted analysis, COVID-19 remained associated with higher risk of short-term mortality (adjusted odds ratio [aOR] 2.54 [95% 2.39-2.70]), with findings on sensitivity analyses consistent with the primary model among sepsis hospitalization subsets with septic shock ([aOR] 2.70 [95% 2.51-2.91]) and ICU admission ([aOR] 2.67 [95% 2.30-3.10]). CONCLUSION(S): COVID-19 infection was associated with over 250% higher odds of short-term mortality among septic patients. Additional studies are needed to determine the mechanisms underlying these observations in order to inform future efforts to reduce the observed outcome disparities.
ABSTRACT
INTRODUCTION: Decreasing case fatality among septic patients has been documented in the United States (US). The strain on healthcare resources brought by the COVID-19 pandemic has been associated with a rise in adverse health outcomes in non-COVID patients. However, the populationlevel impact of the COVID-19 pandemic on the case fatality in sepsis among non-COVID patients is unknown. METHOD(S): We used a statewide dataset to identify hospitalizations aged >=18 years in Texas during April 1-December 31, for each year of 2016-2020 (to align each year with the date of introduction of COVID-19-specific ICD-10 code [U071] in the US). Sepsis was defined by "explicit" ICD-10 codes for severe sepsis (R65.20) and septic shock (R65.21). COVID-19 hospitalizations were excluded. Hierarchical models were fit to estimate the changes in shortterm mortality (defined as in-hospital death or discharge to hospice) of sepsis hospitalizations using 2 approaches: 1) using the 2016-2019 data to forecast risk-adjusted shortterm mortality in 2020 and then comparing the predicted and observed 2020 mortality 2) using the 2019-2020 data to estimate the change in short-term mortality in 2020. RESULT(S): There were 207,953 sepsis hospitalizations without a diagnosis of COVID-19 during the study period (45,826 in 2019 and 41,996 in 2020). Short-term mortality has decreased between 2016 and 2019 from 29.7% to 26.6% (adjusted odds ratio [aOR]/year 0.93 [95% CI 0.92-0.94]). The predicted and observed short-term mortality among sepsis hospitalizations in 2020 was 25.8% (95% CI 25.6-26.0) vs 30.8%, respectively (p < 0.0001). Following adjustment for confounders, the risk of short-term mortality among sepsis hospitalizations was higher in 2020 than in 2019 (aOR 1.30 [95% CI 1.25-1.35]). CONCLUSION(S): The COVID-19 pandemic was associated with reversal of the progressive pre-pandemic downtrend in case fatality of septic patients, with 30% higher odds of short-term mortality in 2020 compared to the preceding year among sepsis hospitalizations without COVID-19. Further studies are needed to determine the patient-, health system-, and policy-related contributors to these findings in order to inform potential scalable strategies to reduce pandemicrelated adverse impact on outcomes of septic patients without COVID-19.
ABSTRACT
INTRODUCTION: Acute respiratory distress syndrome (ARDS) is the major manifestation of severe respiratory failure due to COVID-19 and is present in the majority of COVID-19-related deaths in autopsy studies. Thus, the COVID-19 pandemic is expected to change substantially the epidemiology of ARDS. However, the contribution of COVID-19 to ARDS-related mortality in the United States (US) is unknown. METHOD(S): We used the CDC WONDER Multiple Cause of Death Data set to identify decedents with a diagnosis of ARDS during 2015-2019, and with a diagnosis of COVID-19, ARDS, or both during 2020. ARDS and COVID-19 were identified by ICD-10 codes J80 and J071, respectively. Negative binomial regression was used on the 2015-2019 data to forecast the number of ARDS-related deaths in 2020. We then compared the number of observed vs expected ARDS-related deaths in 2020. In addition, we examined the reporting of a diagnosis of COVID-19 among decedents with ARDS and the proportion of a diagnosis of ARDS among those with COVID-19. The latter analyses were then repeated across the Department of Health and Human Services (HHS) Regions. RESULT(S): In 2020, there were 51,184 ARDS-related deaths, 384,536 COVID-19-related deaths, and 41,606 deaths with both in the US. The predicted number of ARDSrelated deaths for 2020 was 10,851 (95% CI 9,714-12,120). The ratio of the observed vs expected ARDS-related deaths was 4.71 (95% CI 4.62-4.82). A diagnosis of ARDS was reported in 10.8% of all COVID-19 related deaths, ranging from 8.2% (HHS Regions 1 & 7) to 16.1% (HHS Region 2). COVID-19-related deaths have contributed to 81.3% of observed ARDS-related deaths in 2020, varying from 68.8% (HHS Region 10) to 91.5% (HHS Region 2). CONCLUSION(S): The number of ARDS-related deaths in the US increased nearly 5-fold in 2020, due to the contribution of ARDS among COVID-19 decedents. However, ARDS was reported only in about 1 in 10 COVID-19-related deaths, with the frequency of ARDS diagnosis varying nearly 2-fold across HHS Regions. Our findings suggest that the major rise in ARDS-related deaths in the US in 2020 is nevertheless an underestimate of the actual toll of ARDS-related mortality that year, likely reflecting substantial underdocumentation and possibly underrecognition of ARDS among COVID-19 decedents.
ABSTRACT
INTRODUCTION: Rural residence has been associated with increased risk of COVID-19-related mortality. However, the population-level prognostic implications of rural residence among critically ill patients with COVID-19 are lacking, and the impact of inter-hospital transfer and hospitals' location on the outcomes of these patients is unknown. METHOD(S): We used a statewide dataset to identify ICU admissions aged >=18 years with a diagnosis of COVID-19 in Texas during April 1-December 31, 2020. COVID-19 was defined by ICD-10 code U07.1. We used dichotomized (rural vs urban) ZIP Code-level Rural-Urban Commuting Area categories, linked to hospitalization data, to identify rural residence. Hierarchical, mixed-effects models were fit to estimate the association of rural residence with shortterm mortality (defined as in-hospital death or discharge to hospice) for the whole cohort and among hospitalizations with and without transfer from another hospital. Similar modeling was used to examine the association of care in rural hospitals among rural residents without transfer to another facility with short-term mortality. RESULT(S): Among 58,485 ICU admissions with COVID-19, 9,495 (16.2%) were rural residents. Among rural residents, 8,607 (90.6%) were managed in non-rural hospitals, and 1,827 (19.2%) were transferred from another hospital. The unadjusted short-term mortality among rural and urban residents was 25.9% vs 23.9%, respectively. Following adjustment for confounders, rural residence was associated with higher short-term mortality for the whole cohort (adjusted odds ratio [aOR] 1.093 [95% CI 1.003-1.191]) and among those transferred from another hospital (aOR 1.349 [95% CI 1.106-1.646]), but not among those without inter-hospital transfer (aOR 1.052 [95% CI 0.955-1.159]). Management of critically ill rural residents with COVID-19 in rural hospitals, without inter-hospital transfers was not associated with shortterm mortality on adjusted analyses (aOR 0.672 [95% CI 0.393-1.149]). CONCLUSION(S): The observed increased short-term mortality among critically ill patients with COVID-19 residing in rural areas is confounded by inter-hospital transfers and the geographic location of hospitals, with no adverse prognostic impact of rural residence in non-transferred patients and those managed in rural facilities.
ABSTRACT
INTRODUCTION: Recent reports suggest very low to no hospital survival among COVID-19 patients with in-hospital cardiac arrest (IHCA). However, studies to date included generally very small number of IHCA events and were often single-centered. The population-level outcomes of IHCA among COVID-19 patients is unknown. METHOD(S): We used a statewide data set to identify hospitalizations aged >=18 years in acute care hospitals in Texas with a diagnosis of COVID-19 between April 1st and December 31st, 2020. COVID-19 infection was identified using ICD-10 code U071. Cardiopulmonary resuscitation was identified using ICD-10 code 5A12012. Hospitalizations with cardiac arrest as a primary diagnosis and those without a primary diagnosis of COVID-19 were excluded. Mixed-effects multivariable logistic regression modelling was used to identify predictors of hospital survival among those with IHCA. RESULT(S): Among 65,482 hospitalizations with COVID-19, 893 (1.4%) had IHCA. Among those with IHCA, 57.1% were aged >= 65 years, 64.2% male, 70.9% racial/ethnic minority, and 7.1% had shockable rhythm. IHCA occurred in 12.7% [95% CI 11.8-13.6] of terminal hospitalizations. Hospital survival was 7.3% [95%CI 5.6-9.3], ranging from 6.7% [95% CI 4.6-9.3] among those aged >=65 years to 10.7% [95% CI 4.6-21.0] among those aged < 45 years. On adjusted analyses, among examined patient and hospital characteristics, only shockable rhythm (adjusted odds ratio [aOR] 2.63 [95% CI 1.05-6.56]) and management in hospitals with 200-399 beds (aOR 0.14 [95% CI 0.03- 0.58]), but not demographics, comorbidities, or illness severity, were associated with hospital survival. Among hospital survivors, 23.1% were transferred to hospice and 35.4% were discharged home. CONCLUSION(S): Resuscitation of IHCA among COVID-19 patients occurred more selectively compared to the general population. Hospital survival was very low, and less than 3% of those with IHCA were discharged home. Once developing among patients with COVID-19, the short-term survival of IHCA was no longer affected by demographic characteristics, comorbidity burden, or illness severity. Further large studies, using granular data, are needed to better guide clinicians', patients', and surrogates' decision-making and to improve patients' outcomes.
ABSTRACT
INTRODUCTION: The adverse impact of comorbid conditions on the development of severe illness and risk of death among hospitalized patients with COVID-19 has been well-documented. However, the population-level epidemiology and outcomes of previously healthy [PH] adults compared to those with prior comorbidities [PC] among COVID-19 patients requiring ICU admission are unknown. METHOD(S): We used a statewide dataset to identify hospitalizations aged >=18 years with ICU admission and a diagnosis of COVID-19 in Texas during April 1-December 31, 2020. COVID-19 was defined by ICD-10 code U07.1. PH was defined as absence of the comorbidities included in the Charlson Comorbidity Index, and of obesity, malnutrition, mental disorders, and substance and alcohol use disorders. A hierarchical, mixed-effects model was fit to estimate the association of PH with short-term mortality (defined as in-hospital death or discharge to hospice) among ICU admissions. A similar approach was used to identify predictors of short-term mortality among the PH group. RESULT(S): Among 58,845 ICU admissions with COVID-19, 6,760 (11.6%) were PH. Compared to those with PC, those with PH were younger (aged >=65 years 36.1% vs 49.4%), more commonly racial/ethnic minority (63.8% vs 61.5%), and with lower mean [SD] number of organ dysfunctions (1.2 [1.1] vs 1.8 [1.4]) [p< 0.001 for all comparisons]. Short-term mortality was lower among PH than among PC (16.4% vs 25.0%). However, following adjustment for confounders, the risk of short-term mortality was higher among PH (adjusted odds ratio [aOR] 1.37 [95% CI 1.25-1.51]). Among PH ICU admissions, short-term mortality increased with age ([aOR] 35.20 [95% CI 22.09-56.09];>=65 vs 18-44 years) and management at facilities with >=50 ICU beds ([aOR] 4.43 [95% CI 1.07-18.32] vs < 10 ICU beds). CONCLUSION(S): PH was uncommon among critically ill adults with COVID-19 and PH patients had substantially lower short-term mortality than those with PC. However, once risk-adjusted, the odds of short-term mortality were, unexpectedly, 37% higher among PH, with the latter facing higher risk of death when managed at hospitals with higher number of ICU beds. Additional studies are needed to identify the patient-, care process-, and health system-related contributors to these findings.