Basic Demographic Parameters Help Predict Outcomes in Patients Hospitalized With COVID-19 During the First Wave of Infection in West Texas.

The severity of COVID-19 ranges from asymptomatic subclinical infections to severe acute respiratory failure requiring mechanical ventilation. Patients admitted to the hospital have increased mortality rates and patients requiring intensive care have significantly increased mortality rates. Multiple factors influence these outcomes. This study used simple demographic information available on admission to evaluate possible associations between these variables and outcomes, including mortality and length of stay. Clinical outcomes in 63 patients admitted to a tertiary care hospital in West Texas were reviewed. Older patients, patients admitted from nursing homes, and patients admitted to medical intensive care units had increased mortality. Unadjusted analysis indicated that males had increased mortality. Adjusted analysis indicated that males spent nearly 5 days longer in the hospital than females. In summary, age, chronic illness requiring nursing home placement, and acute severe illness requiring intensive care unit admission identify patients with worse prognoses. In addition, males will likely have a longer length of hospital stay.

Introduction

Coronavirus 2 (SARS-CoV-2) has caused a worldwide pandemic. The infectious syndromes range from asymptomatic subclinical infections to severe acute respiratory failure requiring mechanical ventilation. The mortality rates clearly depend on the population analyzed. Chen et al reviewed risk factors for mortality in hospitalized patients in a nationwide study in China. The case mortality rate was 3.14%.[1] Risk factors for mortality included age greater than 75, age between 65 and 74, coronary heart disease, cerebrovascular disease, presentation with dyspnea, elevated procalcitonin levels, and elevated aspartate aminotransferase levels. This information is clearly needed to classify patients according to disease severity, identify comorbidity that may complicate the clinical course, and structure conversations with patients and families about prognoses. In this study we analyzed basic demographic factors available on admission to determine whether or not they were associated with mortality and length of stay. This study depends on outcomes from one center during the initial wave of infection in West Texas. It provides the basis for comparison with future studies that may have different outcomes based on changes in the care of these patients.

Methods

Medical Record Review

A list of patients with COVID-19 infections established by PCR tests was obtained from the Infection Control and Prevention Office at University Medical Center in Lubbock, Texas. The PCR tests used in our hospital include Xpert® Xpress SARS-CoV-2 (Cepheid, Sunnyvale, CA), The BD SARS-CoV-2 Reagents for BD MAX™ System (Becton, Dickinson and Company, Sparks, Maryland), and The DiaSorin Molecular Simplexa™ COVID-19 Direct real-time RT-PCR (DiaSorin Molecular LLC, Cypress, California). This hospital is a 480 bed tertiary care center with a Level 1 trauma designation. The medical intensive care unit has 32 beds and supports a pulmonary and critical care medicine fellowship. The timeframe for hospitalization for these patients ranged from March 1 through May 15, 2020 discharge date. Patients were included if they were deceased or discharged by May 15th. Medical records were reviewed to determine demographic characteristics, symptoms, comorbidity, initial vital signs, initial laboratory tests, initial chest X-ray abnormalities, and initial site of hospitalization (inpatient ward service vs intensive care unit). Outcomes, including mechanical ventilation, continuous renal replacement therapy, length of stay, and mortality, were determined. Although large amounts of clinical data were collected on these patients, the main effort in this report was to evaluate the association of key demographic variables which would be available on admission with outcomes. Power analysis was not conducted as this study was intended as a descriptive summary of all available COVID-19 hospitalizations in the IRB approved time frame.

Statistical Analysis

Data were summarized using means and standard deviations, median and interquartile ranges, and numbers with percentages. Ordinary least squares regression analysis was used to determine the effect of variables on length of stay. Logistic regression was used to determine the association of variables on the binary outcome of inpatient mortality. P values less than .05 were considered significant.

Study Approval

This study was approved by the Institutional Review Board (L20-172) at Texas Tech University Health Sciences Center in Lubbock, Texas, and by administrative review at University Medical Center (UMC) in Lubbock, Texas.

Results

This study included 63 patients with a mean age of 62.1 ± 14.1 years (median age 62 years with an interquartile range of 54 to 72). Thirty-six patients (57.1%) were males. The in-hospital mortality rate was 30.2%. The mean length of stay was 10.2 days. Table 1 demonstrates the final disposition for every patient admitted to UMC with a COVID-19 positive test during this study period. The table is stratified by either MICU or hospital admission. Final dispositions included death, home, home with hospice, long term acute care, nursing home, prison, or in-patient rehabilitation center. The death rate for patients admitted to the MICU was 60%; the death rate for patients admitted to the hospital was 16.3%. Seven patients were discharged and readmitted; 2 of these patients died.

Table 1.

Total patient final disposition.

	Deceased N (%)	Home N (%)	Home with hospice N (%)	LTAC N (%)	Nursing Home N (%)	Prison N (%)	Rehab N (%)
Total 63	19 (30.2%)	28 (44.4%)	1 (1.6%)	1 (1.6%)	7 (11.1%)	2 (3.2%)	5 (7.9%)
MICU admit n = 20	12 (60%)	3 (15%)	1 (5%)	0 (0%)	2 (10%)	2 (10%)	0 (0%)
Hospital admit n = 43	7 (16.3%)	25 (58.1%)	0 (0%)	1 (2.3%)	5 (11.6%)	0 (0%)	5 (11.6%)
MICU only n = 16	12 (75%)	1 (6.3%)	0 (0%)	0 (0%)	1 (6.3%)	2 (12.5%)	0 (0%)
Hospital only N = 28	2 (7.1%)	17 (60.7%)	0 (0%)	0 (0%)	4 (14.3%)	0 (0%)	5 (17.9%)
MICU to hospital N = 4	0 (0%)	2 (50.0%)	1 (25.0%)	0 (0%)	1 (25.0%)	0 (0%)	0 (0%)
Hospital to MICU N = 15	5 (33.3%)	8 (53.3%)	0 (0%)	1 (6.7%)	1 (6.7%)	0 (0%)	0 (0%)

Abbreviation: LTAC, long term acute care; Rehab, rehabilitation center.

Table 2 stratifies the fatality and survival rates by age group. The highest fatality rate was seen in the 80 to 89 age group (60%). The highest survival rate was in the 30 to 39 age group (100%).

Table 2.

Outcome based on age.

Age (Min = 30, Max = 93)	N	Death	Case fatality rate %	Survived	Case survival rate %
0-29 years	0
30-39 years	4	0	0.00%	4	100.00%
40-49 years	7	2	28.6%	5	71.4%
50-59 years	17	4	23.5%	13	76.5%
60-69 years	15	4	26.7%	11	73.3%
70-79 years	13	5	38.5%	8	61.5%
80-89 years	5	3	60.0%	2	40.0%
90+ years	2	1	50.0%	1	50.0%

Abbreviation: N, number.

Table 3 reports patient outcomes by origin at time of admission. Most patients were admitted from their homes (54.0%) or from nursing homes (30.2%). Patients from the nursing homes had the highest mean age (72.0 ± 12.0) and a mortality rate of 47.4%. Patients from the prison had the lowest mean age (56 ± 9.0) and the highest mortality rate (60%).

Table 3.

Outcome based location prior to admission.

Origin location	N (% total)	Mean age	Deaths, N (% for row)	Death mean age
Home	34 (54.0%)	56.26	6 (17.7%)	63.3
Nursing home	19 (30.2%)	71.95	9 (47.4%)	73.3
Rehabilitation center	5 (7.9%)	68.8	1 (20.0%)	67.0
Prison	5 (7.9%)	56.0	3 (60.0%)	50.3

All ages are years.

Table 4 reports patient outcomes and length of stay stratified by either MICU or hospital admission. Both the mean length of stay and mean age for the MICU admissions (10.9 days and 64.6 years) were higher than hospital admissions (9.9 days and 60.4 years). Hospital admissions had higher survival rates (57.1%) than MICU admissions (40.0%). The survival rate for patients transferred from the MICU to the hospital was 100% with a mean LOS of 17.3 days. The survival rate for patients transferred from the hospital to the MICU was 66.7% with a mean LOS of 14.4 days. The Figure 1 plots the hospital location and length of stay for all patients.

Table 4.

Outcomes and length of stay.

	Mean age admit	Mean LOS total	Death (N, %)	Mean age of deceased	Mean LOS of deceased	Survived (N, %)	Mean age of survivors	Mean LOS of survivors
Total patients n = 63	62.1	10.2	19 (30.2%)	67.2	10.6	44 (69.4%)	59.8	10.1
MICU admit n = 20	64.6	10.9	12 (60%)	67.3	9.4	8 (40.0%)	60.5	13.1
Hospital admit n = 43	60.4	9.9	7 (16.3%)	67.1	12.7	36 (83.7)	59.7	9.4
MICU admit n = 20	64.6	10.9	12 (60%)	67.3	9.4	8 (40.0%)	60.5	13.1
MICU only n = 16	64.8	9.3	12 (75%)	67.3	9.4	4 (25%)	57.5	9.0
MICU to hospital n = 4	63.5	17.3	0	NA	NA	4 (100%)	63.5	17.3
Hospital admit n = 43	60.4	9.9	7 (16.3%)	67.1	12.7	36 (57.1%)	59.7	9.4
Hospital only n = 28	59.2	7.5	2 (7.1%)	66.0	9.0	26 (92.9%)	58.7	7.4
Hospital to MICU n = 15	64.1	14.4	5 (33.3%)	67.6	14.2	10 (66.7%)	60.9	14.5

All ages are years.

Abbreviation: LOS, length of stay.

Figure 1.

This figure plots the length of stay in days on the x axis and the patient number on the y axis. The yellow bar indicates hospitalization in the MICU and the blue bar indicates hospitalization in a non-MICU in-patient unit.

Analysis of these demographic factors using ordinary least squares regression indicates that the male gender was associated with a 5-day longer length of stay compared to females (b = 4.78, P = .028) (Table 5). Logistic regression analysis using age and gender alone demonstrated that age was positively and significantly associated with mortality (OR = 1.05, P = .042). However, in the adjusted model adding readmission and place of origin, age no longer had a significant effect (data not shown).

Table 5.

Ordinary least squares regression predicting length of stay.

	b	95% CI	P-value
Male	4.85	0.57, 9.14	.027
Age	0.04	−0.15, 0.22	.692
Readmission	1.89	−5.26, 9.04	.598
Origin (home = ref)
Nursing home	1.34	−4.50, 7.17	.648
Prison	1.64	−6.53, 9.81	.689
Rehabilitation	3.96	−4.12, 12.04	.33
Disposition—deceased	−0.81	−5.53, 3.90	.732
R-squared	0.1207

Discussion

This study demonstrates that older patients, patients admitted to medical intensive care units, and patients admitted from nursing homes have higher mortality rates with COVID-19. Unadjusted analysis suggests that males have higher mortality rates. The sample size was too small to generalize about COVID-19 admissions from prison, but a noticeable characteristic of our sample shows that all 5 such cases in our data were admitted directly to the MICU and 3 of them died in the hospital. Adjusted analysis indicates that maleCOVID-19 patients spend about 5 days longer in the hospital regardless of final disposition. These simple indicators provide the basis for organizing care and developing conversations with both patients and families regarding prognosis. Gold et al have reported that 26.2% of patients hospitalized in 8 Georgia hospitals did not have high risk of medical conditions.[2] Consequently, other factors such as those used in this study could help classify patients into risk categories.

Other studies have used larger databases to identify clinical indicators which are associated with increased mortality. Chen et al reviewed 50 deaths in a total cohort of 1590 cases throughout China.[1] The median age of fatal cases was 69 years; 60% of these cases were male. Based on a multivariate Cox progression analysis, the following factors were independent risk factors for a fatal outcome: age greater than 75, coronary heart disease, cerebrovascular disease, dyspnea, and elevated procalcitonin levels, and elevated aspartate aminotransferase levels. Complications in fatal cases included acute respiratory failure consistent with acute respiratory distress syndrome (ARDS), acute renal failure, secondary infection, septic shock, and disseminated intravascular coagulation. They developed a nomogram which can help identify high risk patients who would require more intensive surveillance and possibly secondary treatment regimens based on complications. Their discussion suggested that this virus binds tightly to the angiotensin-converting enzyme 2 receptor on cell surfaces and that this might explain its increased virulence. Whether or not demographic factors such as age change the number of enzyme receptors or the distribution of enzyme receptors on cell surfaces is unknown.

Price-Haywood analyzed the hospitalization rate and in-hospital mortality rate of patients admitted to an integrated healthcare system in New Orleans.[3] In this study 1382 patients were admitted to the hospital, and 326 patients died, resulting in a mortality rate of 23.6%. In a multivariable model, the risk factors for mortality included age, a high respiratory rate (>24 breaths per minute), absolute lymphocyte count below 1000/µL, a platelet count below 150 000/µL, and a creatinine above 1.5 mg/dL. Race and gender were not factors in mortality. Goyal and colleagues described the clinical characteristics of patients with COVID-19 hospitalized in New York in March 2020.[4] This study included 393 patients with a median age of 60.2 years. Thirty-three percent of these patients required mechanical ventilation. Of those patients requiring mechanical ventilation, 95.4% needed vasopressor support, and the mortality was 14.6%.

Gupta et al reported outcomes on the multicenter study of patients in United States admitted to medical intensive care units.[5] This study included 2215 patients (64.8% males) with mean age of 60.5 ± 14.5 years and at least 1 chronic comorbidity. At 28 days after ICU admission 35.4% of patients had died. Factors independently associated with death included older age (greater than 80 years), male sex, higher body mass index (>40 compared to <25), coronary artery disease, active cancer, hypoxemia, liver dysfunction, and renal dysfunction at the time of ICU admission. In addition, hospitals with fewer ICU beds (<50) had higher mortality rates. There were significant variations among these hospitals in the use of proning and/or neuromuscular blockade during mechanical ventilation and medications, such as hydroxychloroquine and tocilizumab.

Tang and co-authors compared the clinical characteristics of patients hospitalized with ARDS caused by COVID-19 or caused by H1N1 influenza virus.[6] This study included 73 patients with COVID-19 and 75 patients with H1N1 influenza. Based on a multivariable analysis, patients with a COVID-19 infection were more likely to be older, have fatigue, have gastrointestinal symptoms, and present with ground glass opacities on chest computed tomography (CT). Patients with H1N1 influenza infection were more likely to have higher Sequential Organ Failure Assessment scores, sputum production, and consolidation on CT. The in-hospital mortality in patients with a COVID-19 induced ARDS was 28.8%; the in-hospital mortality of patients with H1N1 induced ARDS was 34.7%. In general, patients with COVID-19 infections presented with lower severity of illness scores. This could lead to inappropriate triage of critically ill patients.

The limitations in our study include a small sample size in a single tertiary care center. The characteristics of this population may differ from other sites due to several factors, including severity of outbreak and available resources. Further, these estimates are not necessarily representative of future COVID-19 hospitalizations in the same hospital, as treatments improve and as the demographics of the affected population fluctuate over time. Sample size did not permit measurement of the association of specific comorbid conditions on mortality and length of stay. Individuals hospitalized during the time frame who had not been discharged or who died after May 15th were excluded from the study, and therefore, some outliers may be excluded from the sample. However, the primary effort was to determine whether or not simple demographic parameters would help identify patients with poor outcomes. As expected, nursing home status which likely reflected increased comorbidity, ICU admission which reflected increase severity of disease, and increased age were associated with increased mortality. Male gender was associated with prolonged hospital stays. More complex analysis including documented comorbidity and admission laboratory tests can generate better predictions of risk of mortality, but factoring multiple variables into patient assessment significantly complicates the initial evaluation of patients. Our study confirms current thinking about critical factors in patient outcomes and will allow comparisons with future studies which may have different outcomes as the clinical experience managing these patients increases and as new effective therapies are identified.