New York Inner City Hospital COVID-19 Experience and Current Data: Retrospective Analysis at the Epicenter of the American Coronavirus Outbreak

Introduction

There is an emergent need for more research and data sharing to better delineate and understand the disease process of the coronavirus across the world. The ongoing pandemic has struck monumental changes in every aspect of human life [1]. A large amount of coronavirus cases resulted in a massive outbreak in New York State and New York City, making it the epicenter of the world [2]. Since the outbreak, rapid publications from Europe [3] and China [4] helped the rest of the world understand the disease course and severity at varying degrees. New York is one of the most culturally diverse places in the world with varying demographics across the city. Our institution in Manhattan has been at the forefront in this hotspot combating the coronavirus disease (COVID-19) pandemic at its peak. As there is a severe underrepresentation of American minority and underprivileged communities, we felt the urgent need to research and analyze various parameters associated with the disease among our patients that might help a deeper understanding and lead to further research from our peers across the world in mitigating the COVID-19 pandemic.

Methods

Study Design

Ethical approval from the hospital Institutional Review Board for Health Sciences Research was expeditiously sought and approved. This study entails data collection of all patients that were hospitalized and had a confirmed case of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. 

Data Collection 

Further review using the electronic medical record was undertaken to obtain the parameters included in this study, namely, diagnosis; age; sex; ethnicity; BMI; in-hospital SARS-CoV-2 conversion status (patients without COVID-19–related symptoms upon admission and subsequently tested positive during hospital stay; hospital acquired); need for mechanical ventilation; duration of mechanical ventilation; fever (core body temperature); length of stay; death; comorbidities: hypertension (HTN), chronic kidney disease (CKD), coronary artery disease (CAD), diabetes mellitus (DM), bronchial asthma (BA), and chronic obstructive pulmonary disease (COPD); inflammatory markers: procalcitonin, C-reactive protein (CRP), lactate dehydrogenase (LDH), and ferritin; and laboratory values: creatinine (Cr), blood urea nitrogen (BUN), D-dimers, prothrombin time (PT), partial thromboplastin time, troponin, blood natriuretic peptide (BNP), glucose, electrocardiogram (EKG), chest x-ray (CXR) findings, and treatment received (azithromycin, hydroxychloroquine). 

Inclusion and Exclusion Criteria

Out of 204 total patients, 184 were admitted with high suspicion of coronavirus infection or other diagnoses (who later had a conversion) between March and April 2020 and were included in our study. The included patients were confirmed positive for SARS-CoV-2. Of the 204 patients, 20 were excluded, as they tested negative. 

Statistical Analysis 

Baseline characteristics of all patients with laboratory confirmed SARS-CoV-2 were included in our study and described in detail. Gross numbers with percentages were calculated and used. We summarized the continuous variables using mean and standard deviation, whereas categorical variables were summarized using counts and percentages. Continuous variables were primarily analyzed using the student t test with Welch correction, while categorical variable analysis were accomplished using the Fisher exact test; subgroup analyses were performed using a two-tailed chi-square test. P<.05 was considered statistically significant. A Microsoft Excel (Microsoft Corporation) spreadsheet was used to display extrapolated data. Graph Pad Prism 8 (Graph Pad Software) and R software version 4.0.0 (R Project for Statistical Computing) were used for statistical analysis.

Results

Definitions

Patients included in our study for the purposes of understanding the disease severity and statistical analysis were further classified into the following groups: the entire group (termed total), nonintubated total (NI-T), nonintubated living (NI-L), nonintubated deceased (NI-D), intubated total (I-T), intubated living (I-L), and intubated deceased (I-D). We compared parameters across all patients and groups to delineate deeper clinical understanding with associations as detailed further in the results and Multimedia Appendix 1. 

Patient Characteristics 

The total number of patients included in our study was 184 of the total 204 as detailed in the inclusion and exclusion criteria in the Methods section. All of our patients were SARS-CoV-2 confirmed by polymerase chain reaction of the collected nasopharyngeal sample. 

The overall mortality in our study was 17.39% (n=32). In our study, 16.30% (n=30) required mechanical ventilation, and the mortality rate was 43.33% (n=13/30) among those intubated. Of the 184 patients, 21 (11.41%) had contracted SARS-CoV-2 in the hospital who were otherwise admitted for other reasons, mostly surgical. The average age was 64.72 (range 28-97, SD 14.87) years. Over one-third of the patients included were female and less than two-thirds were male. Mean age with SD for all groups is reported in Table 1. We found that younger patients were more likely to be intubated than their older counterparts. The comparison between NI-T and I-T yielded a P value of .01, and the comparison between NI-L and NI-D yielded a P value of .51, and between I-L and I-D yielded a P value of .60. 

BMI

Of the 184 patients, the mean for BMI was 29.10 (SD 7.386, range 13.38-61.68) kg/m², of which more than one-fourth had a BMI<24.0 kg/m², less than one-tenth had a BMI of 25-29.9 kg/m², less than one-fourth were in class 1, and class 2 and class 3 each had less than one-tenth. We then analyzed to see if the BMI was associated with either mortality or the increased need for mechanical ventilation. We found that BMI had correlations for both NI-T and I-T, with means of 28.56 and 31.74, respectively (P=.07); NI-L and NI-D had means of 29 and 25.71, respectively (P=.04), and I-L and I-D had means of 33.98 and 28.57, respectively (P=.07).

Ethnicity

Of the 184 patients, less than one-third were Hispanic, more than three-fifths were African Americans, less than one-tenth were Asians, and there was 1 Caucasian. For African Americans, the comparison of NI-T and I-T had a P value <.001; I-L and I-D yielded a P value of .05. For Hispanics, a comparison of NI-T and I-T yielded a P value of .02; I-L and I-D had a P value of .07. All subgroup analysis findings are listed in Table 1. 

Comorbidities

The average comorbidity was 1.70 for all patients admitted with SARS-CoV-2 infection. The average comorbidity among patients in each group were NI-T: 1.73; I-T: 1.57; NI-L: 1.64; NI-D: 2.32; I-L: 1.41; and I-D: 1.77. There was statistical significance with a P value of .045 on comparing NI-L and NI-D, clearly indicating that increased comorbidity among patients not requiring mechanical ventilation positively correlated with mortality. We then calculated percent associations for each group based on the presence of existing comorbidities. Overall, almost two-thirds of the overall group had HTN. There was statistical significance when comparing NI-L and NI-D (P=.02), indicative of a positive correlation of HTN with mortality; there was no difference between the I-T and NI-T (P=0.91). Overall, less than one-fifth of the patients had CKD; there was no statistical significance when comparing all 3 groups for CKD as depicted in Table 2. Overall, about one-fifth had CAD. There were no statistical differences noted among the groups for CAD; however, a comparison of NI-L and NI-D yielded a P=.07. Overall, over two-fifths of our patient population had DM. We had statistical significance between the NI-T and I-T for DM (P<.001). Both BA and COPD made up less than one-tenth of our patients. There was no significance among any groups for BA and COPD. Table 2 is representative of the percent associations of various comorbidities with P values for the groups compared.

Inflammatory Markers

We then assessed the distribution of inflammatory markers across various subgroups to gain an understanding of them and to see if this would serve as an indirect surrogate marker of the disease severity. Details of inflammatory markers along with other laboratory values with range, mean, SD, and the standard error of the mean (SEM) for each subgroup are depicted in Table 3 along with the results of the statistical analysis.

Procalcitonin

There was significant statistical difference between the NI-T and I-T groups (95% CI 0.7077-12.04, P=.03); however, we found no difference between NI-D and NI-L (95% CI 1.322-5.607, P=.20) or between I-D and I-L (P=.23).

C-Reactive Protein

Comparison of NI-T and I-T yielded a P value of .006 (95% CI 2.516-14.82), and a comparison of NI-D and NI-L had a P value of .24 (95% CI –21.20 to 64.19). Finally, I-D compared to I-L had a P value of .76 (95% CI –11.77 to 8.871).

Ferritin

There was no statistical significance in all 3 comparison groups, which were as follows: NI-T and I-T (95% CI 1075-1144, P=.95), NI-D and NI-L (95% CI –1538 to 400, P=.24), I-D and I-L (95% CI –1975 to 1040, P=.52).

Lactate Dehydrogenase

We found that there was statistical significance of LDH between NI-T and I-T (95% CI 88.31-318.6, P=.001), while no differences were found between I-D and I-L or the NI-D and NI-L groups (P=.57 and P=.34, respectively). 

Fever

The mean temperature was 100.9° F (SD 1.68, SEM 0.1239) across all patients. Plotted values for each subgroup are shown in Table 1. On comparing the temperature between NI-T and I-T, we found statistical significance (P=.004) with means of 100.7° F and 101.8° F, respectively. There was no significant difference linking temperature to mortality among the remainder of the subgroup analysis.

Prothrombin Time

There was no statistical significance across all three groups for PT: NI-T vs I-T (95% CI –1.1997 to 2.457, P=.49), NI-D vs NI-L (95% CI –13.49 to 18.52, P=.65), and I-D vs I-L (P=.57).

Partial Thromboplastin Time

No statistical difference between all 3 comparison groups with the following: P=.16, P=.67, and P=.84 comparing NI-T vs I-T and then between the NI and I subgroups, respectively. 

D-Dimers

D-dimers were elevated on most of our patients with SARS-CoV-2 confirmed cases. There were no significant differences among the groups; although, the intubated patients had a mean that showed a clear trend of elevation compared to those nonintubated. NI-T vs I-T had a P value of .45, with a mean for NI-T of 4512 and for I-T of 6636. Further analysis between the deceased and living patients of the nonintubated and intubated groups yielded P=.47 and P=.17, respectively. 

Glucose

There was statistical significance between the NI-T and I-T groups, who had mean glucose levels of 218.4 and 297.3, respectively (95% CI 18.46-139.3, P=.01). There was no significance between the nonintubated subgroups NI-D and NI-L (P=.18); however, it is noteworthy that the mean glucose levels among the deceased were 287.2 compared to 218.4 for the living. We noted the same trends among the intubated group with a mean of 354.2 among the deceased vs 253.9 for I-L. The comparison of I-D vs I-L yielded a P value of .07. 

Creatinine

There was a clear trend of increasing Cr values, indicating acute kidney injury (AKI) proportional to the disease severity regardless of their pre-existing renal function status. NI-T vs I-T was statistically significant (P=.03), with a mean Cr of 2.486 and 4.030, respectively (95% CI 0.1641-2.924). Although there was no statistical significance among the NI-D vs NI-L, it is quite clear that up trending Cr was indicative of worsening renal function among the deceased P=.09. The patients intubated overall had worse renal function, but there was no significance between groups (P=.77). 

Blood Urea Nitrogen

We performed the same grouped analysis for BUN. There was significant differences correlating that of Cr among the NI-T and I-T (95% CI 11.54-43.82, P=.001). There was statistical significance among the living and deceased in the nonintubated population (95% CI 1.148-47.85, P=.04). No statistical differences were detected between the I-L and I-D group (P=.56).

Troponin

Although troponin had been elevated, indicating myocardial stress, in most of our patients that had it drawn, a correction was done for those with and without pre-existing CAD and potential influence with coexisting CKD. We did note an elevation across all groups. NI-T and I-T had means of 0.03853 and 0.04862, respectively (P=.72); NI-D vs NI-L had a P value of .28, and I-D vs I-L had a P value of .48. 

Brain Natriuretic Peptide

There was no significance for all three groups. Of note, we had lower sample size and reduced power in the groups after correction; only a few of our patients had BNP drawn. Among those that had it drawn, most of them had their values elevated. NI-T vs I-T had a P value of .12, and NI-D vs NI-L had a P value of .56.

Electrocardiogram

Our EKG findings among patients and their distribution across various subgroups are described in Multimedia Appendix 1. Predominant EKG findings were sinus rhythm (n=94/160, 58.75%); sinus tachycardia (n=34, 21.25%); sinus tachycardia with prolonged QT (n=3, 1.875%); prolonged QT (n=6, 3.75%); and arrhythmia, blocks, or other (n=23, 14.38%). Comparison of NI-T and I-T yielded a P value of .08 for tachycardia; comparison of NI-L and NI-D had a P value of .02 for arrhythmias and heart blocks, indicating that patients with pre-existing arrhythmias or heart blocks had positive associations with mortality. 

Chest X-Ray

We classified CXR findings into 7 major classes as reported by the radiologist for all the patients in the study group. We sorted out associations as depicted in Table 4 with percent distributions, group analysis, and P values. Of significance, we found that NI-T and I-T comparisons were statistically significant (P=.02) for bilateral (B/L) patchy opacities, which was a predominant finding among the nonintubated patients and with B/L diffuse pulmonary infiltrates predominantly in the intubated group, for which the comparison of NIT to I-T was statistically significant (P<.001). 

Treatment

Of the total 184 patients, 153 patients received some form of treatment as categorized in Table 5 of our manuscript. The primary treatment groups were group 1 receiving azithromycin and hydroxychloroquine, group 2 receiving Azithromycin only, group 3 receiving hydroxychloroquine only, and group 4 no treatment. They were not grouped based on any specific parameters, but treatments were given with emerging science and potential alleviation of the disease process. Percent distribution and P values are described in Table 5. We found that no combination of treatment made any difference in preventing the need for mechanical ventilation or mortality. 

Discussion

Our study population was primarily grouped into patients that required admission with and without the need for mechanical ventilation, we further classified and subgrouped them to deceased and living, this renders a deeper perspective on the entire spectrum of COVID-19 among the hospitalized. Our patient population is unique in that it gives a fair representation of the American minority. It was interesting to note that in our study younger patients were more likely to need mechanical ventilation than their older counterparts, while the older patients had increased mortality, and this positively correlated with increasing comorbidity. We also found that the comparatively lower BMI group had worse prognosis in both needing mechanical ventilation and mortality contrary to our traditional beliefs [5]. Increased core body temperature corresponded to worsening disease. Our study had striking differences in that more African Americans with increased comorbidities had higher mortality and decreased need for mechanical ventilation compared to Hispanics, who were more likely to be intubated and then succumb to the disease. HTN was independently associated with mortality among the nonintubated. Inflammatory markers were increased across the board for all of our patients with significant differences among the I-T and NI-T groups and trends between the deceased and living; this certainly proves that they act as surrogate markers of disease severity in SARS-CoV-2 infection. Recent studies linked AKI to COVID-19 [6], our study not only confirmed their findings but also showed directly proportional worsening of renal function with increasing disease severity. Mean blood glucose followed the same trajectory, corresponding to disease severity indicative of insulin resistance insurgence [7]. Pre-existing conduction defects, atrial fibrillation, and arrhythmias positively correlated with mortality among the nonintubated. We also noted that patients with B/L CXR findings of patchy opacities were more likely to be hospitalized without the need for mechanical ventilation and recovery versus those with B/L diffuse infiltrates, who were more likely to be intubated. Our study suggests azithromycin and hydroxychloroquine treatment when given individually or as combination therapy did not alter disease progression and mortality.

Although our study clearly illustrates associations of age, BMI, ethnicity, body temperature, multiple comorbidities, HTN, DM, CAD, inflammatory markers, mean blood glucose, AKI, EKG findings, and CXR findings with likelihood of needing intubation or mortality, it is not without limitations. This is an early retrospective and not a randomized controlled study. Sample size, demographics, and ethnic composite in our study needs to be taken into account while interpreting results. We urge the readers, clinicians, and researchers to take our early findings, bearing in mind its limitations while making clinical decisions. Further investigation and research on a large scale based on our early findings is essential.