Predictors of disease severity in patients hospitalized with coronavirus disease 2019

Abstract

BACKGROUND:

Coronavirus disease 2019 (COVID-19), caused by a novel coronavirus, manifests as a respiratory illness primarily and symptoms range from asymptomatic to severe respiratory syndrome and even death. During the pandemic, due to overcrowding of medical facilities, clinical assessment to triage patients for home care or in-hospital treatment was an essential element of management.

OBJECTIVES:

Study the demographic features, comorbidities and bio-markers that predict severe illness and mortality from COVID-19 infection.

DESIGN:

Retrospective observational

SETTING:

Single tertiary care center

PATIENTS AND METHODS:

The study included all patients admitted with a positive PCR test for COVID-19 during the period from March 2020 to September 2020 (7 months). Data on demographics, clinical data and laboratory parameters was collected from medical records every 3 days during hospital stay or up until transfer to ICU.

MAIN OUTCOME MEASURES:

Demographic, comorbidities and biochemical features that might predict severe COVID-19 disease.

SAMPLE SIZE:

372

RESULTS:

Of the 372 patients, 72 (19.4%) had severe disease requiring admission to intensive care unit (ICU); 6 (1.6%) died. Individuals over 62 years were more likely to be admitted to the ICU (P=.0001, while a BMI of 40 and higher increased the odds of severe disease (P=.032). Male gender (P=.042), hypertension (P=.006) and diabetes (P=.001) conferred a statistically significant increased risk of admission to ICU, while coexisting COPD, and ischemic heart disease did not. Laboratory features related to severe COVID-19 infection were: leukocytosis (P=.015), thrombocytopenia (P=.001), high levels of C-reactive protein (P=.0001), lactic dehydrogenase (P=.0001), D-dimer (P=.0001) and ferritin (P=.001). With the multivariate analysis, diabetes, high lac-tate dehydrogenase, C-reactive protein and thrombocytopenia were associated with severity of illness.

CONCLUSIONS:

Particular demographic and clinical parameters may predict severe illness and need for ICU care.

LIMITATIONS:

Single referral center, several cases of severe COVID-19 could not be included due to lack of consent and or data.

CONFLICT OF INTEREST:

None.

INTRODUCTION

Severe acute respiratory syndrome, caused by a novel coronavirus (SARSCoV-2), results in major morbidity and mortality. The impact of this disease has had catastrophic effects on health care and the economy all over the world. As of 6 October 2022, there were 616 427 419 confirmed cases of COVID-19, including 6 528 557 deaths reported to the WHO (World Health Organization) from across the world.¹ Although there has been a decline in the number of cases of COVID infections, it is likely that a surge in infection can occur as happened in July 2021 with the appearance of the Delta variant. It is not clear yet if the currently circulating new XBB.1.5 variant, nicknamed “Kraken” could drive up infections.

Characteristics of COVID-19 infection range from asymptomatic to symptoms and signs of upper respiratory infection including anosmia (loss of smell), and pneumonia which could progress to severe respiratory distress syndrome, and possibly death. Many also suffer from gastrointestinal symptoms of nausea, diarrhea, and or dysguesia (loss of taste). Early identification of the factors affecting the severity of COVID-19 is vital, as it can help to prevent adverse outcomes by preemptive, evidence-based treatment.^2,3

Based on studies from various countries, the most common factors that predict severe COVID-19 infection are advanced age, male sex, and presence of comorbidities, like cardiovascular disease, diabetes, hypertension or chronic obstructive pulmonary disease (COPD) in addition to high C-reactive protein (CRP), D-dimer, and lymphopenia.^3–7 With COVID-19 infection, around 75% of deaths have occurred in patients with one or more chronic diseases, mainly diabetes and hypertension.^2,5–15 Triage to assess whether disease was suitable for home care or required in-hospital treatment was an essential element of management due to overcrowding of medical facilities. Risk calculators are available online on CDC website (https://www.cdc.gov/) to risk stratify patients with COVID-19 infections.

The objectives of this study were to identify patient characteristics, comorbid conditions, and laboratory markers that could predict impending severe disease, and or death due to SARS-CoV-2 infection in adults.

METHODS

A hybrid retrospective/prospective cohort study was undertaken in a tertiary care hospital, of all consecutive patients, aged 14 years and above, admitted with a positive real-time reverse transcriptase polymerase chain reaction (RT-PCR) test for COVID-19, during the period March 2020 to September 2020 (7 months). The study was conducted after getting approval from the Office of Research Affairs and Ethics Committee of the hospital.

COVID-19 infection was diagnosed if SARS-CoV-2 was detected in a nasopharyngeal sample by a specific RT-PCR test (RealStar SARS-CoV-2 RT-PCR Kit RUO Altona Diagnostics, Germany), performed in the Department of Pathology and Laboratory Medicine at our institution. COVID-19 infection was classified by the hospital COVID-19 committee as 1) asymptomatic (Stage A, no signs or symptoms of infection; 2) mild infection (Stage B, with upper respiratory tract infection symptoms or other mild symptoms, including fever and gastrointestinal symptoms without evidence of pneumonia; 3) moderate infection (stage C, with hypoxia with oxygen saturation less than 93% at rest or presence of pneumonia, not requiring intensive care unit (ICU) admission); or 4) severe infection (stage D, with pneumonia and requiring ICU admission with any of the following: respiratory rate of 30 breaths/minute, arterial oxygen partial pressure to fractional inspiratory oxygen ratio (PaO₂/FiO₂) less than 300, more than 50% lung involvement on imaging within 24-48 hours, critical respiratory failure requiring mechanical ventilation, septic shock or multi-organ dysfunction.

Data on demographics, comorbidities, medications and laboratory parameters was collected from the electronic medical records every 72 hours during hospital stay or up until transfer to ICU. Results of full blood count, the basic metabolic panel, and inflammatory markers were collected to look for a trend that could predict impending severe disease. Data of 372 patients admitted to the COVID wing of the hospital was analyzed. We excluded patients transferred to our hospital from other secondary care facilities, who lacked the initial laboratory data and detailed history.

The software Stata (StataCorp. 2021. Stata Statistical Software: Release 17. College Station, TX: StataCorp LLC.) was used for analyzing the data. Frequencies and percentages were used to summarize and report categorical variables. Odds ratios were calculated to compare categorical variables. Unadjusted logistic regression was used to compare the lab values. Results from the logistic regression are presented as odds ratio (95% confidence interval). Stepwise forward selection was used to perform a multivariable logistic regression model. A P value less than .05 is considered as statistically significant.

RESULTS

Of the 372 hospitalized patients, 72 (19.4%) had severe disease requiring admission to ICU; 6 (1.6%) died (Table 1). Male patients showed a significant increase in risk for serious infection as did increasing age. Patients aged over 62 years showed more than a tenfold increase in progression to severe disease when compared to patients less than 35 years old (P=.00001). BMI of ≥40 increased the odds of severe disease by 2.8 (95% CI-1.0–7.1). Univariate logistic regression analysis revealed the risk was highest in patients with heart failure (Table 2). Furthermore, having type 2 diabetes mellitus and or hypertension doubled the risk of admission to ICU. Being a current or former smoker did not portend severe disease. Laboratory markers that were related to severe COVID-19 infection were a high white cell count (WBC) C-reactive protein (CRP), lactic dehydrogenase (LDH), D-dimer and ferritin (Table 3). With rising levels of each of these markers, the odds of severe illness increased. The most important predictors for ICU admission, using multivariate analysis were diabetes mellitus, LDH, CRP and thrombocytopenia (Table 4).

Table 1.

Demographics.

Characteristics	Number of patients	Admitted to ICU	Not admitted to ICU	Odds ratio (95% Cl)	P value
Sex
Males	187	44 (60.27)	143 (47.83)	1.7 (1.02, 2.91)	.042
Females	185	28 (39.73)	157 (52.17)	Reference
Age (years)					.0001
15–36	93	4 (6.85)	89 (29.53)	Reference
37–52	100	16 (21.92)	84 (28.18)	4.23 (1.36, 13.19)	.013
53–61	88	18 (24.66)	70 (23.49)	5.72 (1.85, 17.67)	.002
≥62	90	34 (46.58)	56 (18.79)	13.50 (4.54, 40.12)	.0001
BMI (mg/kg2)					.156
≤24.9	98	17 (23.29)	73 (24.5)	Reference
25–29.9	128	21 (28.77)	107 (35.91)	0.93 (0.46, 1.88)	.851
30–34.9	79	17 (24.66)	62 (20.47)	1.40 (0.61, 2.76)	.484
35–39.9	39	7 (9.59)	32 (10.74)	1.04 (0.39, 2.75)	.933
≥40	27	10 (13.7)	17 (5.7)	2.80 (1.09, 7.17)	.032
Blood groups					.137
O	142	29 (39.73)	113 (37.92)	Reference
A	74	18 (24.66)	56 (18.79)	1.25 (0.64, 2.44)	.510
B	55	19 (26.03)	36 (12.08)	2.05 (1.03, 4.09)	.040
AB	13	5 (8.22)	8 (2.35)	2.43 (0.74, 8.00)	.142

Data are number (%).

Table 2.

Comorbidities and admission to ICU.

Comorbidities	Total (n)	Admitted to ICU	Not admitted to ICU	Odds ratio (95% Cl)	P value
Total	372	72	300
Hypertension
Yes	143	38 (26.6)	105 (73.4)	2.1 (1.2, 3.5)	.006
No	229	34 (14.8)	195 (85.1)
Diabetes mellitus
Yes	120	35 (29.2)	85 (70.8)	2.4 (1 4 4 0)	.001
No	252	37 (14.7)	215 (85.3)
Heart failure
Yes	23	12 (52.2)	11 (47.8)	5.2 (2.2, 12.5)
No	349	60 (17.2)	289 (82.8)
Ischemic heart disease
Yes	24	8 (33.3)	16 (66.7)	2.2 (0.9, 5.4)	.08
No	348	64 (18.4)	284 (81.6)
Chronic kidney disease
Yes	32	9 (28.1)	23 (71.9)	1.7 (0.8, 3.9)	.193
No	340	63 (18.5)	277 (81.5)
End Stage Renal Disease
Yes	10	1 (10.0)	9 (90.0)	0.5 (0.1, 3.6)	.459
No	362	71 (19.6)	291 (80.4)
Liver disease
Yes	19	4 (21.0)	15 (78.9)	1.1 (0.4 3.5)	.848
No	353	68 (19.3)	285 (80.7)
COPD
Yes	5	2 (40.0)	3 (60.0)	2.83 (0.5, 17.2)	.26
No	367	70 (19.1)	297 (80.9)
Asthma
Yes	40	7 (17.5)	(82.5)	0.9 (0.4, 2.1)	.753
No	332	65 (19.6)	267 (80.4)
Current smoker
Yes	15	3 (20.0)	12 (80.0)	1.0 (0.3, 3.8)	.949
No	357	69 (19.3)	288 (80.7)
Former smoker
Yes	10	1 (10)	9 (90)	0.46 (0.06, 3.65)	.459
No	362	71 (19.61)	291 (80.39)
Active malignancy
Yes	29	7 (24.14)	22 (75.86)	1.36 (0.56, 3.32)	.499
No	343	65 (18.95)	278 (81.05)
Autoimmune disease
Yes	23	5 (21.74)	18 (78.26)	1.17 (0.42, 3.25)	.77
No	348	67 (19.25)	281 (80.75)
Solid organ transplant
Yes	34	4 (11.76)	30 (88.24)	0.52 (.18, 1.55)	.247
No	338	68 (20.12)	270 (79.88)

Data on ICU admission is number (%).

Table 3.

Unadjusted logistic regression for lab results.

Lab variables	N	Admitted to ICU	Odds ratio (95% Cl)	P value
White blood cell counts (109/L)				.05
≤3.5	53	6 (8.2)	Reference
3.6–11	300	58 (79.4)	1.87 (0.76, 4.6)	.169
11.1–15	14	6 (8.2)	5.87 (1.51, 22.82)	.011
≥15.1	5	2 (4.1)	11.75 (1.62, 85.16)	.015
Absolute lymphocyte (109/L)				.001
≤1.5	186	49 (79)	Reference
≥1.6	172	13 (20.9)	0.22 (0.11, 0.43)	.0001
Ferritin level (μg/L)				.001
≤200	150	10 (16.9)	Reference
201–400	69	11 (18.6)	2.65 (1.06, 6.59)	.035
401–800	61	14 (23.7)	4.17 (1.73, 10.01)	.001
≥801	72	24 (40.6)	7 (3.12, 15.69)	.0001
C-reactive protein (mg/L)				.001
≤5	237	15 (27.2)	Reference
≥5.1	77	40 (72.7)	15.99 (8.04, 31.83)	.0001
D-dimer (μg/mL FEU)				.001
≤0.49	181	11 (17.4)	Reference
0.5–0.6	47	9 (14.2)	3.66 (1.41, 9.45)	.007
0.65–1.0	59	19 (30.1)	7.34 (3.23, 16.64)	.0001
1.1–2.1	35	12 (19)	8.06 (3.19, 20.37)	.0001
≥2.2	25	12 (19)	14.26 (5.28, 38.53)	.0001
LDH (U/L)				.001
≤249	177	7 (12.5)	Reference
250–≤350	91	15 (26.7)	4.79 (1.87, 12.23)	.001
≥351–≤500	32	15 (26.7)	21.42 (7.67, 59.8)	.0001
≥501	23	19 (33.9)	115.35 (30.91, 430.44)	.0001
ALT (U/L)				.813
≤45	253	48 (68)	Reference
46–90	80	17 (23.6)	1.12 (0.6, 2.08)	.713
≥90	37	6 (8.3)	0.8 (0.31, 2.03)	.648
Platelets (109/L)				.001
≤200	115	38 (52.7)	Reference
≥200	256	34 (47.2)	0.31 (1.89, 5.47)	.001

Data on ICU admission is number (%).

Table 4.

Multivariable logistic regression.

Characteristics	Odds ratio (95% Cl)	P value
Blood group
O	Reference
A	0.57 (0.13, 2.48)	.45
B	3.24 (0.75, 13.92)	.11
AB	1.84 (0.20, 16.97)	.58
Unknown	0 (0, inf)	.99
Lactate dehydrogenase (U/L)
≤249	Reference
250<350	4.21 (1.14, 15.55)	.03
351≤500	15.73 (3.16, 78.07)	<.01
≥501	402.39 (42.43, 6627.49)	<.01
C-reactive protein (mg/L)
≤5	Reference
>5	8.33 (2.63, 26.33)	<.01
Platelets (1OVL)
<200	Reference
≥200	0.14 (0.04, 0.47)	<.01
Diabetes mellitus	5.95 (1.77, 20.01)	<.01

Model summary measures: −2 Log likelihood =90.883, Cox & Snell R Square =.367, Nagelkerke R Square=.66

DISCUSSION

In this study of 372 patients admitted with confirmed COVID-19 infection, increasing age increased the risk for severe disease with patients aged over 62 years being ten times more likely to develop severe disease and thus ICU admission. Old age has carried a high risk for severe disease in all studies looking at predictors of severe disease.^2,5–9 Like several studies from different countries, including a study from Saudi Arabia, there was a significant increase in critical illness among males in our cohort. A systematic review and meta-analysis which evaluated sex-specific mortality, showed no increase in ICU admissions among males, but men had significantly higher mortality.¹⁶ However a study looking at risk factors for severe SARS-CoV-2 infection after vaccination did not report significant increase of severe disease in males.¹⁷

In our study patients with heart failure were five times more likely to be admitted to intensive care, while coexisting diabetes mellitus, and hypertension doubled the risk of severe COVID-19 infection. Heart failure is reported to increase hospitalization and mortality from COVID-19 infection.^18,19

Abnormal immune regulation in diabetic patients results in a decreased immune response. A study from Riyadh showed the prevalence of diabetes to be high among hospitalized COVID-19 patients.^12,13 While diabetic patients have a higher mortality rate than their non-diabetic counterparts, other factors such as old age, congestive heart failure, smoking, beta-blocker use, presence of bilateral lung infiltrates, elevated creatinine and severe vitamin D deficiency compound the risk of a fatal outcome.^11–13 In our cohort, patients with chronic kidney disease, ischemic heart disease, and COPD had a higher odds of severe illness, although the differences were not statistically significant. Similarly, due to small sample size, immunosuppression had no significant effect on outcomes.

Biochemical markers pointing to severe disease were leukocytosis, lymphopenia, thrombocytopenia, CRP, LDH, D-dimer and ferritin. As the levels of CRP, LDH, D-dimer, ferritin increased, the risk of severe disease also increased. The trend of the total WBC count, absolute lymphocyte count and platelet count also foretold an onset of severe disease.

Inflammatory markers like CRP, ESR, serum ferritin and interleukin-6 levels are elevated with SARS-CoV-2 infection.^2,11 IL-6 may act as a functional mediator to recruit lymphocytes into the lung lesions, which can further secrete IL-6 in a vicious cycle, thus aggravating lung injury and hastening death.⁹ Measurement of IL-6 levels, which is a marker of cytokine storm are not available in our laboratory. The mechanism for lymphocytopenia with COVID-19 infection, as with other RNA viral infections is likely to be multifactorial eg. apoptosis, pyroptosis, autophagy, antibody-dependent cell-mediated cytotoxicity (ADCC) compounded further by excess cytokines, chemokines particularly IL-6, which hinders lymphopoesis.⁴ Evidence from previous studies has suggested that the rate of lymphocyte decline is more rapid in patients with severe disease than that in those with mild or moderate disease.⁶

Studies have reported increased risk for severe COVID-19 disease in pregnant women with perinatal complications like preeclampsia, premature birth, still birth, and other conditions.^20,21 With only thirteen pregnant individuals in our study, we were unable to make a firm conclusion. In several studies, smoking has increased morbidity and mortality from respiratory viral infections like influenza, SARS-CoV-1, MERS-CoV, but the effect of smoking with SARS-CoV-2 infection is uncertain. A meta-analysis from China did not show an increased risk in smokers, whereas a meta-analysis and systematic review by Reddy et al concluded that current smoking enhanced the risk for severe disease.²² We were unable to make an inference on the effect of past or current smoking from our cohort as the number of smokers were very small.

The occurrence of liver injury is not uncommon and the etiopathogenesis in patients with SARS-CoV-2 infection is multifactorial. Just as in pulmonary alveolar cells, cholangiocytes of the liver have angiotensin-2 converting enzyme (ACE-2) receptors, to which the virus attaches to enter the cell. Other reasons are the effects of acute inflammation and release of interleukins and inflammatory cytokines, particularly IL-6, underlying liver disease, drug-induced liver injury and alcohol use.^9,23–25

Contrary to other studies, we did not observe a notable relationship between deranged liver function and severe illness. This could be attributed to the Islamic culture, which discourages alcohol consumption, and the prevalence of obesity. Although obesity is on the brink of being an epidemic in Saudi Arabia, the prevalence of non-alcoholic fatty liver disease (NAFLD) is still low when compared with the global prevalence, which is reported to range between 30% to 38%.^26,27 The prevalence in Saudi Arabia is estimated to be 25.7%.²⁸ Moreover, the number of patients with underlying liver disease in our cohort was small, and the use of antiviral agents against SARSCoV-2 was negligible, at the start of the epidemic.

Currently with vaccination and immunity from previous infections, and less virulent mutant strains circulating, there are fewer constraints on health care facilities globally. Immunity from vaccinations and infections wane with time and immunosuppressed individuals remain at risk for severe disease and death. For example, patients on anti-CD20 therapy have been shown to have severe outcomes even after two doses of vacci-nation as humoral response is suboptimal.²⁹ Hence risk stratification of patients by understanding predictors of severe disease is indispensable to alert and help health care workers in the appropriate management.

Experts predict that the threat of COVID-19 infection is not over, as the virus could mutate to a more infectious, virulent one, while immunity is failing. The recent surge in COVID-19 infections in China and Singapore validate this. The restrictions on entry to facilities, public gatherings and travel have been lifted in Saudi Arabia and with waning immunity, the threat of a surge in infections is to be anticipated. Furthermore a study from Saudi Arabia showed that the uptake of booster doses among 2332 participants was a mere 527 (22.6%).³⁰

Study limitations include the partly retrospective aspect conducted at the start of the epidemic when treatment options were limited. Also, external validity may be limited as the data is from a single referral center. Further studies in the Saudi population are needed to see if the predictors of severity and outcomes are altered, with use of nirmatrelvir/ritonavir (Paxlovid), immunomodulators and monoclonal antibodies in high-risk patients.

In conclusion, individuals over 62 years of age and those with heart failure were at the highest risk for severe disease. Obesity, co-existing diabetes mellitus and hypertension doubled the risk for ICU admissions. Leukocytosis, thrombocytopenia, high levels of CRP, LDH, D-dimer and ferritin were the laboratory markers predicting severe illness from COVID-19 infection.

Funding Statement

None