We selected COVID-19 patients admitted to Renmin Hospital of Wuhan University (Wuhan, Hubei, China) from Jan 15 to March 10, 2020. All enrolled patients were definitively diagnosed with COVID-19 pneumonia according to the World Health Organization interim guidelines (World Health Organization., 2020). Patients with positive result on high-throughput sequencing or real-time reverse-transcriptase-polymerase-chain-reaction (RT-PCR) assay of nasal and pharyngeal swab specimens were defined as a confirmed case of COVID-19. Patients with missing clinical information, age less than 18 years, or who were pregnant were excluded. Patients without data on cardiac biomarkers, including high-sensitivity troponin I (hs-TNI) and NT-proBNP, were excluded. Up to March 20, 2020, we screened 1000 cases of confirmed COVID-19 in Renmin Hospital of Wuhan University (Fig. 1). A total of 453 cases did not have cardiac biomarker records; 89 cases involved pregnant patients; 34 cases involved repeat hospitalization; and 44 cases had incomplete data. Follow-up was completed for 414 patients in the hospital. A total of 361 patients were cured and discharged, 43 patients died in the hospital, and 11 patients were still in the hospital at the end of the study period. All the research procedures followed the STROBE cohort reporting guidelines (von Elm et al., 2007). This research was approved by the Institutional Review Board of Renmin Hospital of Wuhan University (Wuhan, Hubei, China. Ethics approval number: WDRY2020-K108). The requirement for written informed consent was waived because this research was retrospective observation and anonymization.

Data were retrieved from electronic medical records. Demographic and clinical parameters included age, sex, exposure history, complications on admission, initial symptoms and signs, comorbidities, drug treatments, and outcomes. Data for cardiac biomarkers (hs-TNI, and NT-proBNP) upon admission and at days 3, 7, 10, 15, and 20 of hospitalization were also collected. Two investigators reviewed and extracted all data independently. Patients were divided into two groups: COVID-19 pneumonia combined with hypertension or COVID-19 pneumonia without hypertension.

Shock and acute respiratory distress syndrome (ARDS) were defined in accordance with the WHO interim guidelines (WHO, 2020). Acute kidney injury was defined based on the highest serum creatinine level and urine output according to the KDIGO clinical practice guidelines (2012). The chronic kidney disease was defined based on Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) (Levey et al., 2009). Chronic liver disease included hepatitis C, hepatitis B, alcoholism, nonalcoholic fatty liver disease (NAFLD), autoimmune hepatitis, and liver cirrhosis (Bashir et al., 2020). COPD was defined according to the most recent Global initiative for chronic Obstructive Lung Disease (GOLD) strategy for diagnosis (Mirza et al., 2018). The severity of disease was staged according to the guidelines for the diagnosis and treatment of COVID-19 (sixth trial edition) published by the Chinese National Health Commission on February 19, 2020. Normal cases were defined as patient with fever and respiratory symptoms, imaging shows pneumonia. Severe cases were defined as either (i) respiratory rate $>$ 30 breaths/min, (ii) oxygen saturation $\le$ 93%, or (iii) PaO2/FiO2 ratio $\le$ 300 mmHg. Critical cases were defined as including one criterion of the following: shock; respiratory failure requiring mechanical ventilation, or admission to the intensive care unit (ICU).

The diagnosis of hypertension at baseline was based on investigator responses to the relevant questions on the case report form. History of clinically significant hypertension, defined as systolic blood pressure of 140 mmHg or higher, diastolic blood pressure of 90 mmHg or higher, or both and was taking at least 1 antihypertensive medication at the time of admission.

The primary composite outcome was admission to the intensive care unit (ICU), mechanical ventilation, or death. We investigated the association between hs-TNI and NT-proBNP and the primary composite outcome and all deaths.

For categorical variables, proportions were used for continuous variables, and means with standard deviations or medians with quartiles were used to describe baseline characteristics. Differences in baseline characteristics between patients with and without hypertension were tested using Chi-square test for categorical variables and Kruskal-Wallis’s test for continuous variables. Kaplan-Meier curves for all deaths and cumulative incidence curves for the composite outcomes were estimated, and differences between groups were compared using the log-rank test. Cox proportional hazard models were used to compare the risk of the patients according to hypertension status for the primary endpoint, all deaths and six comorbidities. Cox proportional hazard models were also used to compare the risk of hs-TNI and NT-proBNP levels for all deaths. The Cox regression models were adjusted in five models.

Model 1: univariate analysis.

Model 2: age + sex.

Model 3 (to adjust difference in Laboratory examination and treatment): age + sex + alanine aminotransferase (ALT) + aspartate aminotransferase (AST) + blood urea nitrogen (BUN) + serum creatinine (sCR) + Glucocorticoid therapy + continuous kidney replacement therapy (CKRT) + Oxygen administration + Noninvasive ventilation (NIV) + invasive mechanical ventilation (IMV) + extracorporeal membrane oxygenation (ECMO).

Model 4 (to adjust difference in complications): age + sex + diabetes + chronic obstructive pulmonary disease (COPD) + chronic liver disease + chronic kidney disease + cerebrovascular disease + cardiovascular diseases.

Model 5 (to adjust difference in complications, blood pressure grade and angiotensin converting enzyme inhibitor/angiotensin receptor blocker (ACEI/ARB) drug used): age + sex + diabetes + COPD + chronic liver disease + chronic kidney disease + cerebrovascular disease + Blood pressure grade + ACEI/ARB.

The upper reference limit (URL) of hs-TNI, defined as the 99th percentile in healthy individuals, is 26 ng/L (26 pg/mL) (Jarolim, 2015). Thus, hs-TNI was dichotomized at or above/below the threshold, and NT-proBNP was also divided by tertiles (upper 75% in the hypertension group).

The study population included 414 hospitalized patients with confirmed COVID-19. The median age was 61 years (IQR 19), and 207 (50.1%) were men. Of these patients, 149 (36.0%) had hypertension (28 combined with chronic stable coronary heart disease, 1 with a history of myocardial infarction, 1 with a history of viral myocarditis, 2 with a history of dilated cardiomyopathy, and 5 with arrhythmia [2 with sick sinus syndrome and with pacemaker implantation; 1 with paroxysmal atrial fibrillation and 2 with bradycardia]). In non-hypertension group, 7 combined with chronic stable coronary heart disease, 3 combined with paroxysmal atrial fibrillation. Patients with hypertension were older and had a longer time from the onset of symptoms to hospital admission than those without hypertension. A larger proportion of patients with hypertension had a history of diabetes and chronic liver disease than those without hypertension. The sex ratio and the history of COPD, cerebrovascular disease, and chronic kidney disease were not different between the two groups (Table 1). The initial symptoms and signs were not significantly different between the two groups; fever, cough, fatigue, and chest tightness/shortness of breath were the most common symptoms, presenting in 73.2%, 47.3%, 20.3%, and 18.1%, respectively, of all patients. The distribution of disease severity differed significantly in patients with and without hypertension. Most patients had moderate disease, but critical disease was more common in individuals with hypertension. Patients with hypertension had a higher rate of complications, including shock, ARDS, and acute kidney injury (AKI) than patients without hypertension. NIV and IMV were more common in individuals with hypertension than in those without hypertension. Oxygen administration, antiviral therapy, glucocorticoid therapy, CKRT, antibiotic use, and ECMO use were not different between the two groups. Higher death rates, and vasopressor use was also observed in the hypertension group than in the nonhypertension group. Furthermore, vasopressor use differences mainly existed in critical type patients (Table 1).

On admission, patients with hypertension had a lower hemoglobin level and higher D-dimer, blood urea nitrogen, and serum creatinine levels than those without hypertension (Table 2). In addition, 243 nonhypertensive patients and 140 hypertensive patients had procalcitonin results. However, there were more cases of procalcitonin $>$ 0.5 ng/mL in individuals without hypertension than in those with hypertension. Leukocyte counts, lymphocyte counts, platelet counts, prothrombin time, activated partial thromboplastin time, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin, albumin, creatinine kinase, lactose dehydrogenase, and CD3, CD4, and CD8 T lymphocyte counts were no significant difference between patients with and without hypertension (Table 2).

Fig. 2.

Mortality and cumulative incidence of composite outcomes in COVID-19 patients with and without hypertension. (A) Kaplan-Meier curves for all deaths in the hospital. (B) Kaplan-Meier curves for the cumulative incidence of the composite outcome (intensive care unit (ICU) admission, mechanical ventilation, or death).

Among the 414 patients, 361 patients were cured and discharged, 43 patients died in the hospital, and 11 patients were still in the hospital at the end of the study period. The percentage of patients admitted to the ICU requiring invasive ventilation and who eventually died was 19.5% (29) in patients with hypertension and 7.9% (21) in patients without hypertension (P = 0.001). The death rates in patients with and without hypertension were 16.8% and 6.8%, respectively (Table 1, P = 0.001). The Kaplan-Meier curves for all deaths and cumulative incidence curves are shown in Fig. 2.

The results of the univariate Cox regression risk model are shown in Table S1. Patients with hypertension had a significantly higher risk of all deaths than nonhypertension patients [HR: 2.68; 95% CI (1.46~4.91)]. Patients with cerebrovascular disease had a significantly higher risk of the all deaths than noncerebrovascular disease patients, even after adjusting for age and sex [HR: 4.98; 95% CI (2.079~11.942)]. Other comorbidities (diabetes, COPD, chronic liver disease, and chronic kidney disease) as well as BP grade did not seem to be high risk factors for death (Table S1). When we adjusted for age and sex, no difference was showed about the risk factor of all deaths between hypertension and nonhypertension patients [HR: 1.77; 95% CI (0.93~3.36)] (Table S1). In patients with hypertension, the use of ACEIs/ARBs was not a risk factor for all deaths (Table S2), even in the univariate Cox model.

Fifteen patients in the nonhypertension group and 4 patients in the hypertension group did not have hs-TNI admission data. A total of 22.1% of patients with hypertension had an hs-TNI value $\ge$ 26 pg/mL, compared to 6.4% of those without hypertension (Table 3). The distribution of standardized (log${}_{10}$) plasma baseline hs-TNI concentrations in patients with and without hypertension is shown in Fig. S1. The frequency distribution curve was shifted to the right in patients with hypertension compared to that of patients without hypertension (Fig. S1A). During hospital admission, the hypertension group had more cases showing an hs -TNI value $\ge$ 26 pg/mL than the nonhypertension group (Table 3). We also analyzed the variance curve of hs-TNI, as shown in Fig. S1B.The median hs-TNI increased gradually, peaked at 7 days and then dropped between 7-20 days after admission in patients with hypertension. The median hs-TNI levels were quite smooth in patients without hypertension and stabilized at the 6 pg/mL level during the 0-20 days follow-up after admission.

Thirty-three patients in the nonhypertension group and 8 patients in the hypertension group did not have NT-proBNP admission data. Patients with hypertension had higher median levels of NT-proBNP at baseline than patients without hypertension [141.9 pg/mL, IQR (50~474) vs. 77.3 pg/mL, IQR (30~219); P = 0.000]. Approximately 13.3% of all patients had a NT-proBNP value in the top tertile ($\ge$ 474 pg/mL; IQR- tertile 4 in the hypertension group) (Table 3). When adjusted for COVID-19 severity, normal and severe type patients with hypertension had more cases showing an hs-TNI value $\ge$ 26 pg/mL or NT-proBNP in the top tertile than nonhypertensive patients. In critical type patients, the number of cases with hs-TNI values $\ge$ 26 pg/mL or NT-proBNP in the top tertile did not differ between patients with and without hypertension (Table 3). The distribution of standardized (log${}_{10}$) median NT-proBNP baseline levels in patients with and without hypertension is shown in Fig. S1. The frequency distribution curve for NT-proBNP was also shifted to the right in patients with hypertension compared to that of patients without hypertension (Fig. S1C). During hospitalization, more patients in the hypertension group had a NT-proBNP value in the top tertile than those without hypertension (Table 3). The variance curve of NT-proBNP, as shown in Fig. S1D, was consistent with the hs-TNI trend curve. The median NT-proBNP increased gradually, peaked at 7 days and then dropped between 7-20 days after admission in patients with hypertension. However, the median NT-proBNP increased until 3 days and then dropped 3~20 days after admission in patients without hypertension.

During follow-up, patients with hs-TNI $\ge$ 26 pg/mL had a significantly higher risk of death irrespective of hypertension status (P 0.05) (Fig. S2A, B). The elevated risk related to hs-TNI persisted in the Cox regression model 1 and model 2 incorporating hs-TNI in a categorical manner ($\ge$ 26 vs. 26 pg/mL) or as a continuous variable (per log${}_{10}$ increase in hs-TNI) in patients with and without hypertension (Table 4). In a multivariate Cox regression model including NT-proBNP (model 3), elevated risk related to hs-TNI $\ge$ 26 pg/mL was only observed in the nonhypertension group. When adjusting for all comorbidities (model 4), elevated risk related to hs-TNI $\ge$ 26 pg/mL was vanished (Table 4). In addition, hypertensive patients with dynamic hs-TNI changes were at higher risk of death than those without dynamic hs-TNI changes (P 0.05) (Fig. S2E).

Patients with NT-proBNP in the highest tertile had a significantly higher risk of death irrespective of hypertension status (Fig. S2C and S2D). In the Cox regression models incorporating NT-proBNP as tertiles or as a continuous variable (per log${}_{10}$ increase in NT-proBNP), the elevated risk related to NT-proBNP persisted in model 1 and model 2. In a multivariate Cox regression model including hs-TNI, the elevated risk related to NT-proBNP as tertiles only persisted in patients with hypertension (model 3, Table 4). When adjusting for all comorbidities (model 4), blood pressure grade and ACEI/ARB usage (model 5), elevated risk related to NT-proBNP at the top tertile persisted in patients both with and without hypertension.