key: cord-0973777-ona4mg5w authors: Antos, A.; Kwong, M. L.; Balmorez, T.; Villanueva, A.; Murakami, S. title: Unusually high risks of COVID-19 mortality with age-related comorbidities: An adjusted method to improve the risk assessment of mortality using the comorbid mortality data. date: 2021-05-21 journal: nan DOI: 10.1101/2021.05.20.21257550 sha: 9f0a1ecf6315c69e835dfad98ae9f368ca294096 doc_id: 973777 cord_uid: ona4mg5w Background: The pandemic of Coronavirus Disease 2019 (COVID-19) has been a threat to global health in the world. In the US, the Centers for Disease Control and Prevention (CDC) has listed 12 comorbidities within the first tier that increase with the risk of severe illness from COVID-19, including the comorbidities that are common with increasing age (referred to as age-related comorbidities) and other comorbidities. However, the current method compares a population with and without a particular disease (or disorder), which may result in a bias on the results. Thus, comorbidity risks of COVID-19 mortality may be underestimated. Objective: To re-evaluate the mortality data from US States and estimate the odds ratios of death by major comorbidities with COVID-19, we incorporated the control population with no comorbidity reported and assessed the risk of COVID-19 mortality with comorbidity. Methods: We collected all the comorbidity data from the Public Health websites of fifty US States and Washington DC accessed on December 2020. The timing of the data collection should allow minimizing a bias from the COVID-19 vaccines and new COVID-19 variants. The comorbidity demographic data were extracted from the State Public Health data made available online. Using the inverse-variance random-effects model, we performed a comparative analysis and estimated the odds ratio of deaths by COVID-19 with pre-existing comorbidities. Results: A total of 39,451 COVID-19 deaths were identified from four States that had comorbidity data, including Alabama, Louisiana, Mississippi, New York. 92.8% of the COVID-19 deaths were associated with pre-existing comorbidity. The risk of mortality associated with at least one comorbidity combined was 1,113 times higher than that with no comorbidity. The comparative analysis identified nine comorbidities with odds ratios of up to 35 times significantly higher than no comorbidities. Of them, the top four comorbidities were: hypertension (odds ratio 34.73; 95% CI 3.63-331.91; p = 0.002), diabetes (odds ratio 20.16; 95% CI 5.55-73.18; p < 0.00001), cardiovascular disease (odds ratio 18.91; 95% CI 2.88-124.38; p = 0.002); and chronic kidney disease (odds ratio 12.34; 95% CI 9.90-15.39; p < 0.00001). Interestingly, lung disease added only a modest increase in risk (odds ratio 6.69; 95% CI 1.06-42.26; p < 0.00001). Conclusion: The aforementioned comorbidities show surprisingly high risks of COVID-19 mortality when compared to the population with no comorbidity. Major comorbidities were enriched with pre-existing comorbidities that are common with increasing age (cardiovascular disease, diabetes, and hypertension). The COVID-19 deaths were mostly associated with at least one comorbidity, which may be a source of the bias leading to the underestimation of the mortality risks previously reported. Taken together, this type of study is useful to approximate the risks, which most likely provide an updated awareness of age-related comorbidities. 3 0.00001), cardiovascular disease (odds ratio 18.91; ; p = 0.002); and 41 chronic kidney disease (odds ratio 12.34; 95% ; p < 0.00001). Interestingly, lung 42 disease added only a modest increase in risk (odds ratio 6.69; 95% CI 1.06-42.26; p < 43 0.00001). Conclusion: The aforementioned comorbidities show surprisingly high risks of 45 mortality when compared to the population with no comorbidity. Major comorbidities were 46 enriched with pre-existing comorbidities that are common with increasing age (cardiovascular 47 disease, diabetes, and hypertension). The COVID-19 deaths were mostly associated with at 48 least one comorbidity, which may be a source of the bias leading to the underestimation of the 49 mortality risks previously reported. Taken together, this type of study is useful to approximate 50 the risks, which most likely provide an updated awareness of age-related comorbidities. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 INTRODUCTION 52 COVID-19 is an infectious disease caused by Severe Acute Respiratory Syndrome We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 77 Protocols (PRISMA-P) statement [5] . The US comorbidity data was searched on the PubMed, We found data from a study conducted in the State of Georgia, but this study did not list the 84 deaths by comorbidity and thus was not included in this study. The Alabama, Louisiana, Mississippi, and New York state department websites update 86 the death tolls for their respective states. We referenced the Alabama [7] , Mississippi [8] , and 87 New York data [9] which were updated on 12/29/2020. We retrieved Louisiana's data on 88 12/17/2020; the Louisiana data were dated on 3/31/2020 [10] . The data surrounding deaths due 89 to COVID-19 in Alabama, Louisiana, Mississippi, and New York were separated and additionally 90 summed for analysis. Deaths that included comorbidity were separated by their respective state. Deaths that did not suffer from comorbidity were also separated by state. To estimate the 92 number of comorbidities in Louisiana, we multiplied the percentages with the total number of 93 deaths since their data only provided a comorbidity percentage. We were unable to input the 94 decimal values from the percentage calculations for Louisiana's numerical fields. Thus, we 95 rounded up if the tenths decimal place was equal to or greater than five and rounded down if the 96 tenths decimal place was less than five. For New York, we also subtracted the total deaths with 97 comorbidity from the total deaths from COVID-19 to deduce how many COVID-19 deaths did 98 not have comorbidity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 To calculate the odds ratios, we used the software, RevMan5 (Cochrane reviews) [11] . The random-effects method was also used to calculate the odds ratios. The inverse variance inverse variance method is so named because the weight given to each study is chosen to be the inverse of the variance of the effect estimate (i.e. one over the square of its standard error). studies, which have larger standard errors. The inverse-variance random-effects model has 106 been commonly used in the COVID-19 meta-analysis studies, for example, [12] and [13] . This 107 choice minimizes the imprecision (uncertainty of the pooled effect estimate)." Statistical 108 heterogeneity is measured as I²>0. The following interpretations for different heterogeneity 109 percentages were given:0% to 40% "might not be important", 30% to 60% "may represent 110 moderate heterogeneity", 50% to 90% "may represent substantial heterogeneity", and 75% to factors include the "magnitude and direction of effects" and "the strength of evidence for 114 heterogeneity (e.g., P-value from the Chi-squared test, or a confidence interval for I²)." We organized the ten comorbidities between the states to standardize them for analysis 116 as follows. Hypertension was documented in Mississippi and New York that were included to 117 estimate odds ratios of hypertension. For cardiovascular disease (CVD), New York listed 118 detailed diseases with CVD pathologies (atrial fibrillation, coronary artery disease, and stroke 119 comorbidity) [9] and thus were included under the term CVD. Three states showed the data for 120 the immunocompromised conditions, liver disease, and neurological disease odds ratios; and 121 two states for the chronic kidney disease and obesity odds ratio figures. For chronic kidney 122 diseases (CKD) and renal disease (RD), CKD is a part of Renal Disease (RD), and thus RD 123 data include CKD; nonetheless, we did perform the analysis of CKD. The WHO designates 124 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 21, 2021. ; https://doi.org/10.1101/2021.05.20.21257550 doi: medRxiv preprint stroke as a neurological disorder [14] and the CDC designates it as cardiovascular or 125 cerebrovascular disease [3] . We followed the CDC designation and thus included stroke data 126 into the cardiovascular disease odds ratios. New York had more detailed designations than 127 other States and we treated the New York data as follows. Dementia data were included into 128 Neurologic Disease, and COPD was included in Lung Disease. In this study, Neurological 129 disease was called Neurologic disease for the consistency of disease designations. Despite the 130 designations by the CDC, the four states also did not distinguish the diabetes types (type 1, type 131 2, other types of diabetes) and is shown as diabetes. We modified the assessment procedure of the COVID-19 mortality data and used the 134 control population with no comorbidity (Methods). We focused on mortality data, since the 135 outcomes from the COVID-19 test results could be influenced by different testing systems and 136 by how they were administered and recorded. For the reason, we excluded the COVID-19 test 137 results. Similarly, we excluded hospitalization that may contribute to a bias of the risk 138 assessment. The US comorbidity data was searched on the PubMed, UpToDate, and Scopus 139 websites. The results from queries were reviewed and data regarding deaths from patients with 140 and without comorbidity were not found. Thus, the CDC and State websites were searched for 141 provisional comorbidity data. The CDC had a provisional dataset of "comorbidities and other 142 conditions"; the dataset was dominated by "pneumonia and influenza", "respiratory failure" and 143 "hypertensive diseases" [15] which were mixed with COVID-19 symptoms, causes of deaths 144 and pre-existing conditions. Thus, we did not use the CDC dataset. We collected all the data available as of December 2020 by searching the Public Health 146 websites of fifty States and Washington DC in the US. We have compiled and organized the ten 147 comorbidities between the states to standardize them for analysis ( Table 1 ). The demography is 148 summarized in Supplementary Material 2. More details are described in the Section of Method. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 21, 2021. ; https://doi.org/10.1101/2021.05.20.21257550 doi: medRxiv preprint study, which may create a bias by dominating the result. To minimize the bias, we used the 154 random-effects models to control the results of the odds ratio. As a comparison, New York 155 accounted for 75.4% of the total deaths in the study while Mississippi accounted for 11.9%. Alabama accounted for 12%, while Louisiana accounted for 0.6% of the total deaths in the 157 study. Therefore, the random-effects model helped mitigate some of the bias from New York. 92.8% of the COVID-19 mortality was seen in patients with at least one comorbidity. 159 Table 1 and Figure 2 summarize the results. Nine out of ten comorbidities combined to 160 show a surprisingly high risk of death due to covid-19, 1,113 times higher (odds ratio, 1113.59; 161 95% CI, 157.59, 7888.28; p <.00001). Of all the comorbidities, the top comorbidities were: 162 cardiovascular disease; chronic kidney disease, diabetes; and hypertension. Nine comorbidities, 163 except chronic kidney disease, showed a significant amount of heterogeneity (I² value ranging 164 from 77% to 100%) present which was supported by a strong heterogeneity p-value (ranging 165 from 0.00001 to 0.01) and high variability in the confidence interval (Table 1) . As shown in Figure 3A , cardiovascular disease has a high risk of death due to covid-19 167 (odds ratio 18.91; 95% CI 2.88-124.38; p = 0.002). Hypertension is often coupled with 168 cardiovascular disease, and when we combined both hypertension and cardiovascular disease, 169 the risk was doubled (odds ratio, 40.70; 95% CI 27.06, 61.21; p < 0.00001). Figure 3B shows 170 the chronic kidney disease (CKD) comorbidity data (odds ratio 12.34; 95% CI 9.90-15.39; p < 171 0.00001). Importantly, no heterogeneity (I² = 0%) was present in CKD which is supported by the 172 weak P-value (P = 0.58) and significant overlap of confidence intervals between the two 173 studies. Figure 3C shows the diabetes comorbidity data (odds ratio 20.16; 95% CI 5.55-73.18; p 174 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 21, 2021. ; https://doi.org/10.1101/2021.05.20.21257550 doi: medRxiv preprint < 0.00001). Figure 3D shows the hypertension analysis containing comorbidity data (odds ratio 175 34.73; 95% CI 3.63-331.91; p = 0.002). A significant amount of heterogeneity (I² = 99%) was 176 present in this analysis which was supported by a strong P-value (P< 0.00001) and no 177 significant overlap of the confidence intervals between the two studies. 178 Figure 3E shows the immunocompromised condition comorbidity data (odds ratio, 6.89; 179 95% CI 3.89, 12.20; p < 0.00001); Figure 3F shows the liver disease comorbidity data (odds 180 ratio, 2.04;95% CI 3.89, 12.20; p = 0.02); and Figure 3G shows the lung disease comorbidity available from the US, we estimated the odds ratios of comorbidities within the US population to 198 assess their impact on mortality from COVID-19. We found age-related comorbidities to be the 199 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 21, 2021. ; https://doi.org/10.1101/2021.05.20.21257550 doi: medRxiv preprint major conditions: cardiovascular disease, chronic kidney disease, diabetes, and hypertension. When cardiovascular disease and hypertension were combined, the odds ratio of mortality 201 became 40 times higher than without comorbidity. The risks of mortality from COVID-19 in 202 patients with medical comorbidities appears to be much higher than previous studies that used 203 non-US data. For example, the CDC refers to the studies about hypertension consist most if not 204 all of non-US data (cdc.gov accessed on March 2021). The odds ratios were much higher in the 205 US (odds ratio 34.7; this study) compared to those in studies conducted in non-US-countries 206 (odds ratios 2.3-6.5 in non-US countries) [e.g., 17-20] , which implies that there seems to be a 207 variation among countries. Thus, a more comprehensive study remains to be done with data 208 from more US states once they become available. We organized the comorbidity categories with consideration given to the varied 210 definitions of categories between organizations and state reports. In the study, we followed the 211 CDC categories whenever possible. Adjustments of the categories for the purposes of analysis 212 in this study are described in the Result and Method sections. We also noted that the CDC has 213 a detailed list of diseases contributing to the cause of death, which is dominated by "pneumonia 214 and influenza" and "respiratory failure" in the category [15] . In our analysis, we used pre-existing the inverse variance method of data analysis to ensure that the data from New York did not 217 overpower data from the other states. Despite the adjustments, the odds ratios were surprisingly 218 higher than those estimated based on previous research 219 Most of the comorbidities showed significantly higher mortality as measured by the odds 220 ratio. Comorbidities that significantly increase the risk of a severe COVID-19 include 221 cardiovascular disease, chronic kidney disease, diabetes, hypertension, immunocompromised 222 condition, liver disease, lung disease, obesity, and renal disease. We could not find a significant 223 increase in risk due to neurological conditions. Heterogeneity was significantly high for nine 224 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 21, 2021. comorbidities, except for chronic kidney disease. Interestingly, dementia, a part of the 225 neurological conditions, was placed in fourth for comorbidities in New York, while other states 226 did not have this category. Therefore, it remains contested whether or not dementia shows a 227 significant increase in mortality with COVID-19. We used an adjusted method that has strengths as follows. Firstly, it uses the control 229 population with no comorbidities which avoid bias from the population. The control is particularly 230 important when most of the COVID-19 deaths are associated with one or more comorbidities. Secondly, the mortality data with comorbidities are readily available to the public, which are 232 expected to be less biased. The sample sizes are similar to the previous systematic review and 233 meta-analysis studies [4, [18] [19] [20] [21] . Thirdly, it is relatively simple to perform the assessment 234 independent of other factors that are expected to have high variability. Finally, we designed the from a public health perspective. Nine out of ten comorbidities analyzed were shown to increase 250 the likelihood of death from COVID-19. We found that in the US population, the odds ratio for 251 each of these comorbid conditions was higher than that previously seen in the studies 252 predominantly from outside of the US. We call for a standardized format and distribution of data 253 be provided by each State regarding COVID-19 morbidity and mortality so that a thorough and 254 cohesive analysis of data can occur. We also propose to include a control population with no 255 comorbidity when possible to minimize a bias from other health conditions. Altogether, this study gives an awareness about the comorbidities that are more deleterious than previously 257 recognized and provides a call to action for the public health community. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 21, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. "Total" columns in the model represent the numerical amount of the total deaths from COVID-356 19. The "Events" column on the left represents the number of deaths that included the 357 cardiovascular disease comorbidity in that respective state. The "Events" column on the right 358 represents the number of deaths without comorbidity in that respective state. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 21, 2021. ; https://doi.org/10.1101/2021.05.20.21257550 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 21, 2021. ; Are patients with hypertension and 260 diabetes mellitus at increased risk for COVID-19 infection? The Lancet Respiratory 261 An interactive web-based dashboard to track COVID-264 19 in real time The centers for disease control and prevention (CDC) The centers for disease control and prevention (CDC) Associated with High Risk for Severe COVID-19: Information for Healthcare Providers 272 Preferred Reporting Items for Systematic Review and Meta-Analysis 276 What are neurological disorders and how 308 many people are affected by them? (n.d.) Weekly Updates by Select 311 Demographic and Geographic Characteristics AlzGene Database: Benefits and Limitations of Using C. elegans for the Study of 315 Alzheimer's Disease and Co-morbid Conditions Evidence-Based Genetics 318 and Identification of Key Human Alzheimer's Disease Alleles with Co-morbidities The Relationship of COVID-19 Severity with Cardiovascular Disease and Its Traditional Risk Factors: A Systematic Review and 323 Decreased Mortality of COVID-19 With Renin-Figure 1 COVID-19 Deaths from four states Alabama: 4,737 deaths Louisiana: 239 deaths Mississippi: 4,719 deaths New York: 29,756 deaths With cardiovascular disease: 2,368 With diabetes With obesity: 60 With chronic kidney disease: 55 With cardiovascular disease: 50 With pulmonary disease: 29 With neurological disease: 14 With immunocompromised condition: 10 With chronic liver disease: 5 No comorbidity: 7 With hypertension: 1,878 With cardiovascular disease: 1,388 With diabetes: 1,149 With obesity: 649 With neurologic conditions: 681 With lung disease: 679 With renal disease With coronary artery disease: 3,638 With renal disease With COPD: 3,034 With atrial fibrillation: 2,496 With cancer: 2,405 With stroke: 1,904 No comorbidity Abbreviations: NA (Not available), NYC (New York City), NYS (New York State). * The Alabama percentages by race add up more than 100%. It is a possibility of patients in two or more race. ** Alabama and Mississippi label Hispanic as an ethnicity, but not a race. Alabama reports 109 of the total deaths were of Hispanic ethnicity and Mississippi reports 50 deaths were of Hispanic ethnicity. *** NYC data and NYS excluding NYC data were available as percentages shown. The numbers were not available.