Keywords biomarkers - albumin - C-reactive protein - SARS-CoV-2 - coronavirus disease 2019
- blood urea nitrogen
Introduction
Since coronavirus disease 2019 was first identified in December 2019, many researchers
around the world have investigated the diagnosis, treatment, and prognostic factors
of the disease.[1 ] The rapidly increasing number of patients placing an additional burden on the health
system has made it necessary to predict the prognosis of this previously unknown disease.
The limited number of hospital beds has increased the importance of identifying the
critical patient and identifying patients in need of medical support.[2 ] Researchers are particularly working on easily accessible and cheaper biomarkers
to detect critical illness.
Hematological parameters and their ratios are the most studied biomarkers for prediction
of mortality or severe illness.[3 ] To increase the predictability of such tests, researchers have also examined the
ratios of these parameters to each other.[3 ]
[4 ]
[5 ] The neutrophil/lymphocyte ratio has been reported to predict severe disease, intensive
care need, and mortality among patients infected with SARS-CoV-2.[5 ] We hypothesized that the ratio of biochemical markers divided by albumin might be
more reliable in predicting short-term mortality in these patients. Since albumin
is a negative acute phase reactant and low albumin levels are associated with mortality
in SARS-COV-2-infected patients, we think that dividing biochemical markers by albumin
will increase the predictability of these parameters. In the current literature, blood
urea nitrogen (BUN)/albumin ratio, lactate/albumin ratio, and C-reactive protein (CRP)/albumin
ratio were tested as a predictor in SARS-CoV-2 infection.
In this study, we sought to compare the predictive power of the BUN/albumin ratio,
lactate/albumin ratio, and CRP/albumin ratio, evaluated at the time of admission,
in predicting all-cause 30-day mortality in SARS-CoV-2-infected patients presenting
to the emergency department (ED) and reveal which is the best predictor.
Materials and Methods
Study Design
This retrospectively designed, single-center, observational study was performed at
the Ümraniye Training and Research Hospital, University of Health Sciences, a tertiary
education health care center with annual ED visits of 438,000 patients and a capacity
of 682 beds. We documented the data of patients admitted to ED with a SARS-CoV-2 infection
between September 1, 2020, and January 1, 2021.
Study Population
Our study population consisted of patients with a SARS-CoV-2 infection confirmed by
the reverse transcription polymerase chain reaction (rt-PCR) test, who presented to
our ED between September 1, 2020, and January 1, 2021. All the hospitalized patients
and outpatients who had one positive result among the several rt-PCR tests performed
were included in the study. Patients who had not been tested for BUN, lactate, CRP,
and albumin were excluded from the study. Patients with chronic renal failure (since
high baseline BUN levels) and chronic liver failure (since there might be weak CRP
response and low baseline albumin values) were also excluded. The flowchart of the
study is shown in [Fig. 1 ].
Fig. 1 Flowchart of the study. CRP, C-reactive protein; SARS-CoV-2, severe acute respiratory
syndrome coronavirus 2.
Data Collection
The data of patients that presented to our ED and had at least one positive rt-PCR
result for SARS-CoV-2 were obtained from the hospital computer-based information system.
The comorbidities, laboratory parameters, and death information of the patients were
recorded. The comorbidities were noted as chronic obstructive pulmonary diseases,
hypertension, diabetes mellitus, coronary artery disease, congestive heart failure,
history of malignancy, hyperlipidemia, and history of cerebrovascular disease. Patients
with chronic kidney disease and chronic liver disease were excluded. White blood cell
count, neutrophil count, lymphocyte count, platelet count, mean platelet volume, plateletcrit,
BUN, lactate, CRP, and albumin were recorded. The all-cause 30-day mortality data
of all patients were obtained using the national death notification system, where
death information for all patients is legally accessible.
Statistical Analysis
SPSS version 22.0 for Windows (SPSS Inc, Chicago, Illinois, United States) was used
to perform all statistical analyses. The normality analysis of data was conducted
using the Kolmogorov–Smirnov's test. Categorical data were compared using the chi-square
test. Quantitative variables were presented as interquartile range (25th–75th percentiles)
and median values. Quantitative variables were compared using the Mann–Whitney's test
or Student's t -test according to the normality of distribution for the two groups. The Bonferroni
correction was used as a method to counteract the problem of multiple comparisons.
To identify variables associated with 30-day mortality status, the univariate analysis
was undertaken using the chi-square test, Fisher's exact test, Student's t -test, and Mann–Whitney's U test, where appropriate. The receiver operating characteristic (ROC) curves were
used to assess the accuracy of parameters to predict mortality, and the results were
reported as the area under the curve (AUC) values. Youden's index was obtained to
determine the optimal cutoff value for scores with the highest sensitivity and specificity.
The AUC, 95% confidence interval (CI), cutoff, sensitivity, specificity, positive
predictive value, and negative predictive value of all parameters are presented in
[Table 1 ]. The AUC values more than 0.7 were accepted as predictive of mortality.[6 ] The comparison of the AUC values was made with the DeLong's equality test. We have
prepared a two-by-two frequency table according to the optimal cutoff values of ratios
determined according to Youden's index and mortality data. Odds ratios (OR) were calculated
by using two-by-two frequency tables of ratios.[7 ] ORs were used to compare the ability of the parameters to predict mortality. Statistical
significance was defined at p < 0.05.
Table 1
Accuracy of ratios measured at the time of emergency department admission in predicting
30-day all-cause mortality
AUC
Cutoff
Sensitivity
Specificity
PPV
NPV
LR+
LR−
p -Value
C-reactive protein/albumin ratio
0.749
≥0.049
92.86
41.62
33.23
94.90
1.59
0.17
<0.001
Blood urea nitrogen/albumin ratio
0.725
≥1.17
66.96
71.23
42.13
87.33
2.33
0.46
<0.001
Lactate/albumin ratio
0.641
≥0.046
56.25
66.48
34.43
82.93
1.68
0.66
<0.001
Abbreviations: AUC, area under the curve; LR, Likelihood ratio; NPV, negative predictive
value; PPV, positive predictive value.
Ethics
Ethical approval was obtained from the local clinical research ethics committee with
the approval number B.10. 1.TKH.4.34. H. GP.0.01/17. We retrospectively reviewed the
data extracted from the computer-based hospital information management system. The
extracted data were solely clinical and did not include any personal or identifiable
information. Therefore, the necessity for informed consent was waived.
Results
Of the 470 patients included in the study, 232 (49.4%) were female. The median age
of the 470 patients was 69 (25th–75th percentiles: 54.75–79) years. A total of 112
patients died within 30 days of ED presentation. The mortality rate of all-cause 30-day
mortality was 23.8% for our study cohort. [Table 2 ] presents the demographic characteristics, comorbid diseases, and laboratory parameters
and the comparison of these variables between the survivor and nonsurvivor groups.
Table 2
Baseline characteristics and laboratory parameters of the enrolled patients and their
comparison between the survivor and nonsurvivor groups
Variables
Total
n = 470
Survivor
n = 358 (76.2%)
Nonsurvivor
n = 112 (23.8%)
p -Values
n (%)/median (25th–75th percentiles)
n (%)/median (25th–75th percentiles)
n (%)/median (25th–75th percentiles)
Age
69 (54.75–79)
64.5 (50–75.25)
79 (70.25–86)
<0.001
< 65 y
195 (41.5%)
179 (50%)
16 (14.3%)
<0.001
≥65 y
275 (58.5%)
179 (50%9
96 (85.7%)
Gender
0.421
Female
238 (50.6%)
185 (51.7%)
53 (47.3%)
Male
232 (49.4%)
173 (48.3%)
59 (52.7%)
Comorbidities
Chronic obstructive pulmonary disease
61 (13%)
41 (67.2%)
20 (32.8%)
0.078
Hypertension
261(55.5%)
183 (70.1%)
78 (29.9%)
<0.001
Diabetes mellitus
151 (32.1%)
106 (70.2%)
45 (29.8%)
0.037
Coronary artery disease
120 (25.5)
80 (66.7%)
40 (33.3%)
0.005
Congestive heart failure
41 (8.7%)
25 (61%)
16 (39%)
0.017
History of malignancy
33 (7%)
22 (66.7%)
11 (33.3%)
0.184
Hyperlipidemia
162 (34.5%)
108 (66.7%)
54 (33.3%)
<0.001
History of cerebrovascular disease
42 (8.9%)
24 (57.1%)
18 (42.9%)
0.002
Laboratory parameters
White blood cell count (/µL)
6.24 (4.77–8.08)
6.04 (4.75–7.71)
7.39 (4.85–10.13)
<0.001
Neutrophil count (/µL)
4.27 (3.12–6.17)
4.02 (2.96–5.82)
5.56 (3.67–8.00)
<0.001
Lymphocyte count (/µL)
1.17 (0.83–1.65)
1.24 (0.89–1.75)
0.92 (0.67–1.34)
<0.001
Platelet count (103/µL)
191.5 (154–247)
196 (156–149)
184.5 (148.5–227)
0.156
Mean platelet volume (fL)
9.75 (9.00–10.60)
9.70 (9.00–10.63)
9.85 (9.00–10.60)
0.718
Plateletcrit (%)
0.19 (0.15–0.249
0.19 (0.16–0.24)
0.18 (0.14–0.24)
0.291
Blood urea nitrogen (mg/dL)
38.52 (27.82–55.64)
34.24 (25.68–47.62)
53.50 (36.38–32.93)
<0.001
Lactate (mmol/L)
1.6 (1.2–2.0)
1.5 (1.1–2)
1.8 (1.3–2.3)
0.002
C-reactive protein, (mg/dL)
4.3 (1.1–9.2)
3.3 (0.7–7.2)
8.8 (4.3–16.2)
<0.001
Albumin (g/dL)
37.8 (34.6–40.7)
38.6 (37.6–41.4)
35.2 (32.8–37.7)
<0.001
Neutrophil/lymphocyte ratio
3.32 (2.27–6.46)
2.96 (2.10–5.38)
5.38 (2.93–9.46)
<0.001
Platelet/lymphocyte ratio
161.44 (116.20–234.39)
154.85 (11.55–220.88)
195.53 (128.82–125.07)
<0.001
C-reactive protein/albumin ratio
0.112 (0.026–0.266)
0.080 (0.018–0.197)
0.262 (0.112–0.470)
<0.001
Blood urea nitrogen/albumin ratio
1.02 (0.70–1.59)
0.92 (0.65–1.34)
1.49 (0.98–2.48)
<0.001
Lactate/albumin ratio
0.041 (0.031–0.055)
0.040 (0.031–0.051)
0.049 (0.037–0.067)
<0.001
Note: Bonferroni-corrected p -value: 0.0019.
The ROC curve analysis was performed to determine the discriminative ability of laboratory
parameters for 30-day mortality. [Table 1 ] and [Fig. 2 ] present the cutoff values of BUN/albumin ratio, CRP/albumin ratio, and lactate/albumin
ratio according to the best Youden's index, as well as their sensitivity, specificity,
likelihood ratio, and AUC values. According to the best Youden's index, the cutoff
and AUC values were determined as 0.049 (sensitivity: 92.86%, specificity: 41.62%)
and 0.749, respectively, for CRP/albumin ratio; 1.17 (sensitivity: 66.96%, specificity:
71.23%) and 0.725, respectively, for BUN/albumin ratio; and 0.046 (sensitivity: 56.25%,
specificity: 66.48%) and 0.641, respectively, for lactate/albumin ratio. There was
no statistically significant difference between the AUC values of CRP/albumin ratio
and BUN/albumin ratio (0.749 and 0.725, respectively, p = 0.425, DeLong's equality test). The AUC values of CRP/albumin ratio and BUN/albumin
ratio (0.749 and 0.725, respectively) were significantly higher than the AUC value
of lactate/albumin ratio (0.641) (p = 0.002 for CRP/albumin ratio vs. lactate/albumin ratio and p = 0.021 for BUN/albumin ratio vs. lactate/albumin ratio, DeLong's equality test).
Fig. 2 Receiver operating characteristic (ROC) curves for the neutrophil/lymphocyte ratio,
platelet/lymphocyte ratio, C-reactive protein (CRP)/albumin ratio, blood urea nitrogen
(BUN)/albumin ratio, and lactate/albumin ratio for the prediction of 30-day mortality
in patients with severe acute respiratory syndrome coronavirus 2 infection.
The OR values of BUN/albumin ratio (≥1.17), CRP/albumin ratio (≥0.049), and lactate/albumin
ratio (≥0.046) for 30-day mortality were determined as 4.886 (95% CI: 3.108–7.682),
9.268 (95% CI: 4.381–19.605), and 2.518 (95% CI: 1.634–3.882), respectively.
Discussion
Many scoring systems and laboratory parameters have been studied to assess mortality
and severity in SARS-CoV-2-infected patients. Identifying an ideal predictor is necessary
for the optimum use of the facilities of hospitals and the health system. In the current
study, we compared three ratios formed with albumin in predicting 30-day mortality.
Our analyses determined that although CRP/albumin ratio had the highest sensitivity
and negative predictive value and BUN/albumin ratio had the highest specificity, BUN/albumin
ratio and CRP/albumin ratio had similar ability to predict all-cause 30-day mortality
in patients with SARS-CoV-2 infection, while lactate/albumin ratio did not have predictive
power. To our knowledge, our study is the first to compare the ability of BUN/albumin
ratio, CRP/albumin ratio, and lactate/albumin ratio to predict short-term mortality
in patients presenting to the ED with a SARS-CoV-2 infection.
The predictive power of BUN/albumin ratio has been investigated in many different
patient groups presenting to ED.[8 ] Since the beginning of the pandemic, researchers have conducted studies revealing
the importance of BUN/albumin ratio in patients with SARS-CoV-2 infection.[9 ]
[10 ] In a study conducted in ED with 600 patients, Küçükceran et al tested the predictive
ability of BUN/albumin ratio for in-hospital mortality. They reported that BUN/albumin
ratio could be used as a predictive parameter with an AUC value of 0.809 and OR value
of 10.44.[9 ] In another study, Huang et al tested the predictive power of BUN/albumin ratio for
severe disease in SARS-CoV-2 infection. The authors reported that an elevated BUN/albumin
ratio level at admission was an independent risk factor for severe disease and could
be used as a predictor with an AUC value of 0.821.[10 ] The results of our study, in line with the literature, revealed that BUN/albumin
ratio might be a predictor of all-cause 30-day mortality in ED with highest specificity.
CRP/albumin ratio has been investigated as a predictor in many diseases,[11 ] and similar to BUN/albumin ratio, it has been determined as a parameter associated
with mortality in SARS-CoV-2-infected patients in the current literature.[12 ]
[13 ] Güney et al demonstrated that BUN/albumin ratio is an independent predictor of in-hospital
mortality in patients hospitalized due to SARS-CoV-2 infection.[12 ] Torun et al showed that CRP/albumin ratio, fibrinogen/albumin ratio, and neutrophil/lymphocyte
ratio were predictors for severe disease in SARS-CoV-2-infected patients. In addition,
they reported that CRP/albumin ratio was the best predictor of severe SARS-CoV-2 infection
when compared with fibrinogen/albumin ratio and neutrophil/lymphocyte ratio, based
on the results of their cohort.[13 ] Demir et al reported a correlation between CRP/albumin ratio and computed tomography
findings in patients with a SARS-CoV-2 infection confirmed by an rt-PCR test.[14 ] Similar to these studies, the analysis results of our study also revealed a relationship
between CRP/albumin ratio and mortality, suggesting that CRP/albumin ratio could be
used as a predictor of mortality in SARS-CoV-2-infected patients with highest sensitivity
and negative predictive value.
Although lactate/albumin ratio is a well-known predictor in critical and septic patients,[15 ] it is the least researched ratio for SARS-CoV-2-infected patients among the ratios
investigated in our study. The Korean Shock Society investigators evaluated the prognostic
value of lactate/albumin ratio in patients with critical sepsis in a multicenter study
and found it to be a good predictor of critical illness with a low AUC value of 0.68.[16 ] Gök et al, in an intensive care study in which they investigated the relationship
between lactate/albumin ratio and short-term mortality, showed that lactate/albumin
ratio could be used as a prognostic factor in SARS-CoV-2-infected patients with an
AUC value of 0.824.[17 ] Although there was a difference between the mortality and survivor groups in terms
of lactate/albumin ratio as a result of the univariate test, lactate/albumin ratio
had the lowest AUC value of 0.641 in the ROC analysis, indicating that it could not
be accepted as a good predictor.[6 ]
The first and main limitation of our study was that it was designed retrospectively.
There may have been other risk factors that could not be measured due to the retrospective
design. In addition, we only included patients who had been tested for parameters
that we investigated. Patients who were not tested by the clinician for these parameters,
especially blood gas analysis for lactate, were not included in the study. This may
have caused our study population to include more severe patients. Furthermore, there
was a significant difference between the deceased and survivor groups in our sample
in terms of age, hypertension, and hyperlipidemia. Although we do not think that this
affected our results, we recommend repeating the study with more homogeneous groups.
Finally, our study had a single-center design, and therefore, our results cannot be
generalized to other health care institutions.
Conclusion
In conclusion, the results of our study show that BUN/albumin ratio and CRP/albumin
ratio can be used to predict 30-day mortality in SARS-CoV-2-infected patients admitted
to ED. Furthermore, CRP/albumin ratio had the highest sensitivity and negative predictive
value, while BUN/albumin ratio had the highest specificity.
