|Year : 2022 | Volume
| Issue : 4 | Page : 371-376
Study of early warning efficiency of different laboratory markers in predicting progression of disease in COVID-19
Vishwanath Krishnamurthy, K Mohammed Suhail, S Shaikh Mohammed Aslam, Madhu P Raj, Prashanth Patil, Priyanka Phaniraj
Department of Internal Medicine, M.S. Ramaiah Medical College and Hospital, Bengaluru, Karnataka, India
|Date of Submission||10-Jun-2021|
|Date of Decision||22-Nov-2021|
|Date of Acceptance||21-Dec-2021|
|Date of Web Publication||16-Nov-2022|
New BEL Road, M.S. Ramaiah Nagar, MSRIT Post, Bengaluru - 560 054, Karnataka
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Introduction: COVID 19 pandemic has given rise to several challenges to clinicians and one of the keys in this is to predict the set of patients who progress from mild disease to moderate and severe. Apart from the symptomatology and signs, there are several lab parameters varying from biochemical, hematological to radiological parameters which help us in stratifying the stage of disease and also in deciding on which set of patients need close and vigilant monitoring. This would help us in better stratification of disease and utilize the available infrastructure and resources in an optimum way for better management of the disease. Aim: To analyze the early warning efficiency of laboratory parameters individually or in combination in predicting the progress of disease in patients from mild to moderate/severe disease. Materials and Methods: This was taken up as a retrospective study with 100 cases and 100 controls. The demographic details, inflammatory markers, biochemical markers and hematological markers were analyzed. Test of significance was employed to compare categorical variables while student t-test was employed to test the difference in the mean value such as age between case and control (Mann–Whitney U-test in parameters not having normal distribution). Receiver operating characteristic (ROC) curve was constructed for these parameters using cases and controls and area under the curve (AUC) were estimated which was used as an indicator of sensitivity and specificity of the parameter in their early warning efficiency. The critical values for each of the parameters either individually or in combination was estimated as well. Results: Among the parameters C reactive protein (CRP), d-dimers and eosinopenia have the best early warning efficiency. The area under the ROCs curve for neutrophil lymphocyte ratio (NLR), CRP. Ferritin, lactate dehydrogenase, Eosinopenia was 0.609, 0.947, 0.614, 0.554, 0.617 respectively at triage. However, a combination of eosinopenia with CRP (AUC-0.732) or NLR with CRP (AUC-0.728) have a good sensitivity and specificity in predicting the outcome regarding the progression of the disease. Conclusions: Among the parameters, CRP, d-dimers, Eosinopenia and NLR have the best early warning efficiency. However, a combination of Eosinopenia and CRP at triage should also serve as a red flag sign in patients apart from the well-known NLR and IL6 values.
| Abstract in French|| |
Introduction: La pandémie covide 19 a relevé plusieurs défis aux cliniciens et l'une des clés dans ce domaine est de prédire l'ensemble des patients qui passent d'une maladie légère à modérée et sévère. Outre la symptomatologie et les signes, plusieurs paramètres de laboratoire variant des paramètres biochimiques, hématologiques à radiologiques qui nous aident à stratifier le stade de la maladie et également à décider quel ensemble de patients nécessite une surveillance étroite et vigilante. Cela nous aiderait à mieux stratification des maladies et à utiliser l'infrastructure et les ressources disponibles de manière optimale pour une meilleure prise en charge de la maladie. Objectif: Analyser l'efficacité d'alerte précoce des paramètres de laboratoire individuellement ou en combinaison pour prédire les progrès des maladies chez les patients d'une maladie légère à modérée / sévère. Matériaux et méthodes: Ceci a été considéré comme une étude rétrospective avec 100 cas et 100 contrôles. Les détails démographiques, les marqueurs inflammatoires, les marqueurs biochimiques et les marqueurs hématologiques ont été analysés. Le test de signification a été utilisé pour comparer les variables catégorielles tandis que le test T des étudiants a été utilisé pour tester la différence de valeur moyenne telle que l'âge entre le cas et le contrôle (test U Mann - Whitney dans les paramètres n'ayant pas de distribution normale). La courbe des caractéristiques de fonctionnement du récepteur (ROC) a été construite pour ces paramètres en utilisant les cas et les contrôles et la zone sous la courbe (AUC) ont été estimés qui ont été utilisés comme indicateur de sensibilité et de spécificité du paramètre dans leur efficacité d'alerte précoce. Les valeurs critiques pour chacun des paramètres individuellement ou en combinaison ont également été estimées. Résultats: Parmi les paramètres C Protein réactif (CRP), les D - dimères et l'éosinopénie ont la meilleure efficacité d'alerte précoce. La zone sous la courbe ROCS pour le rapport lymphocyte des neutrophiles (NLR), CRP. La ferritine, la lactate déshydrogénase, l'éosinopénie était de 0,609, 0,947, 0,614, 0,554, 0,617 respectivement au triage. Cependant, une combinaison d'éosinopénie avec CRP (AUC - 0,732) ou NLR avec CRP (AUC - 0,728) a une bonne sensibilité et spécificité pour prédire le résultat concernant la progression de la maladie. Conclusions: Parmi les paramètres, le CRP, les D - dimères, l'éosinopénie et le NLR ont la meilleure efficacité d'alerte précoce. Cependant, une combinaison d'éosinopénie et de CRP au triage devrait également servir de signe du drapeau rouge chez les patients en dehors des valeurs NLR et IL6 bien connues.
Mots-clés: C Protéine réactive, efficacité d'alerte précoce, éosinopénie, progression de la maladie dans Covid – 19
Keywords: C reactive protein, early warning efficiency, Eosinopenia, progression of disease in COVID-19
|How to cite this article:|
Krishnamurthy V, Suhail K M, Aslam S S, Raj MP, Patil P, Phaniraj P. Study of early warning efficiency of different laboratory markers in predicting progression of disease in COVID-19. Ann Afr Med 2022;21:371-6
|How to cite this URL:|
Krishnamurthy V, Suhail K M, Aslam S S, Raj MP, Patil P, Phaniraj P. Study of early warning efficiency of different laboratory markers in predicting progression of disease in COVID-19. Ann Afr Med [serial online] 2022 [cited 2022 Dec 9];21:371-6. Available from: https://www.annalsafrmed.org/text.asp?2022/21/4/371/361247
| Introduction|| |
COVID 19 pandemic has given rise to several challenges to clinicians and one of the keys in this is to predict the set of patients who progress from mild disease to moderate and severe. This decision making has become all the more relevant with increased number of home isolation of COVID 19 positive patients who have mild stage of disease on initial triage. Apart from the symptomatology and signs, there are several lab parameters varying from biochemical, hematological to radiological parameters which help us in stratifying the stage of disease and also on deciding on which set of patients need close and vigilant monitoring (e.g., in patients with lymphopenia, raised C reactive protein [CRP] etc.). But most often it is found that each of these different markers/parameters may not be reflecting the same stage of disease especially in the initial stage of the disease. With this background it is of importance to analyze the early warning efficiency of these parameters individually or in combination in predicting the progress of disease in mild patients. This would help us in better stratification of disease and also utilize the available infrastructure and resources in an optimum way for better management of the disease. This study is taken up to understand the early warning efficiency of these markers in mild patients who may progress to moderate/severe disease.
| Materials and Methods|| |
This was a retrospective study on COVID 19 patients admitted to Medical College Hospital in South India. This study consisted of a total of 200 hundred subjects. Equal number of cases and controls were taken for this study. The patients would be classified as mild disease after triaging and looking into the clinicolaboratory parameters as per the MOHFW, Govt of India. The age limit for the inclusion of subjects in the study was a minimum of 18 years and above. As for the study “cases” were defined as patients above the age of 18 admitted to Medical College Hospital who were initially classified as mild COVID disease after triaging and admitted to general wards (non-high-dependency unit [HDU] beds), but over the course of hospitlization, thy progressed to severe disease prompting shift from general wards to intensive care unit (ICU)/HDU. Controls were defined as patients above 18 years of age admitted to Medical College Hospital who were diagnosed with mild COVID disease after triaging and remained as mild disease till recovery without progression to severe COVID disease. The patients' data was retrieved from the medical records department of the Medical College Hospital.
The basic demographic details such as name, age, sex, locality along with the inpatient ID and the status of various comorbidities was noted. Further, three sets of markers were looked into i.e., inflammatory (CRP, ferritin and erythrocyte sedimentation rate [ESR],), biochemical (D-dimer, lactate dehydrogenase [LDH], liver function test, troponins), hematological (complete blood count [CBC], platelet count, neutrophil lymphocyte ratio [NLR] ratio, prothrombin time, activated partial thromboplastin time). The extracted data was tabulated in a MS excel worksheet which was analyzed using SPSS 2009. PASW statistics for windows version, 18.0. Chicago, Illinois, USA.
All the quantitative variables like age, lab values etc., were expressed as descriptive statistics such as mean, median with interquartile range. All the categorical variables such as various comorbidities were expressed in terms of percentage. Test of significance was employed to compare categorical variables while Student's t-test will be employed to test the difference in the mean value such as age between case and control. This is however keeping in mind normal distribution showing a Gaussian curve. In case of parameters such as LDH, IL-6 and D dimer, where the distribution curve was not normal, Mann–Whitney U-test was employed to determine the statistical significance. Receiver operator characteristics curve was constructed for comparing the cases and controls and represent the diagnostic ability of these parameters either individually or in combination. From the Receiver operating characteristic (ROC) curve, area under the curve (AUC) was estimated as a joint indicator of sensitivity and specificity of the parameter. The critical values for each of the parameters either individually or in combination was estimated as well.
| Results|| |
The demographic and clinico-laboratory parameters were compared in the cases group and the control group.
Cases and controls were compared for age, gender and comorbidities. The mean age of patients with COVID 19 was 59.93 and among the controls group was 58.61 (P = 0.07). Among cases, 80.5% of the patients were males and 19.5% females while in controls 76% were males and 24% females (P (males) = 0.18, P (females) = 0.66). The comorbidities were also compared between cases and controls and it was found that both were comparable across age, gender and comorbidities [Table 1].
In further results Laboratory parameters done at triage were compared among cases and controls. Among these parameters were D dimer, NLR, CRP, eosinopenia, LDH, interleukin 6 (IL-6), platelet count, INR, serum ferritin and liver transaminases which are aspartate transaminase (AST) and alanine transaminase (ALT).
[Table 2] shows the mean levels of the lab parameters among cases and controls. The mean level of D-dimers in cases was 1.70 pg/mL as compared to controls was 0.59 pg/mL (P <0.005). Mean value of NLR among the cases was 7.47 and 4.05 in the controls (P < 0.05). The levels of CRP averaged at 79.10 mg/L among cases and in the controls group was 5.89 mg/L (P < 0.05). Above values were statistically significant. Mean level of eosinophils on differential count was 0.92% among the cases and 1.49% among the controls which was statistically not significant (P = 0.148). Mean serum LDH levels were 377.01 U/L among cases and in control was 301.41 (P < 0.05). Mean levels of IL-6 among cases and controls was found to be 75.52 pg/mL and 9.08 pg/mL respectively and difference was not significant (P = 0.175). However IL-6 levels were not sent as a routine marker in many cases and controls hence may not be possible to analyze it. Platelet counts were 2.09 lakhs/cu mm among cases and 2.73 lakhs/cu mm in controls (P = 0.106). The mean value of INR among cases was 1.27 and 1.08 among controls which can be clinically interpreted as elevated. This result though was found statistically not significant (P = 0.202). Average serum ferritin in the cases was 384.29 ng/mL and in the controls was 331.94 ng/mL (P < 0.05). Liver enzymes when estimated and compared among the 2 groups gave the following results. Mean AST level was 53.75 U/L in cases and 52.58 U/L in controls (P < 0.05). Similarly, the mean ALT level was 37.06 U/L in cases and 32.56 U/L in controls (P < 0.5) [Table 2].
|Table 2: Mean values of laboratory parameters amongst cases and controls at triage|
Click here to view
The different lab values at day 0 were analyzed among cases on three different quartiles coarsely representing the mild/moderate/severe disease pattern as depicted in [Table 3]. Seventy five percentage of the patients who progressed to moderate/severe disease had high NLR on day zero. Eighty two percentage of the patients in the cases group had high CRP. Serum ferritin levels were elevated in 29% of the cases on day 0. LDH on day 0 was raised in 57% of the patients who progressed to moderate/severe disease. AST values of 58 case could be retrieved and among them 57 had high AST levels on day 0. Among the 100 cases, 56% of the patients showed mild elevation in serum transaminases, 1% showed elevation beyond 5 times upper limit of normal and 1% showed elevation beyond 10 times the upper limit of normal. Twenty four percentage of the patients among the cases showed thrombocytopenia on day 0. However, 7% of the patients among the 100 had a platelet count of <1 lakh cells/cu mm. Eosinopeneia (defined as AEC <100) was significantly seen in 82% of cases and 48% of controls.
Seeing the significance of eosinopenia as an important marker apart from the well-known NLR and CRP A 2 × 2 contingency table was constructed for eosinopenia amongst cases and controls [Table 4]. Odds ratio was calculated to be 4.94. Relative risk was 2.45 implying that a patient with a mild COVID disease has 2.45 times the chance of progressing into moderate/severe disease if he/she has eosinopenia.
Analysis by receiver operating characteristic
ROC was constructed using cases and controls [Figure 1] and AUC was calculated, further critical values for each of these parameters were derived from the above for these different lab values.
|Figure 1: Receiver operating characteristic curve for various lab parameters|
Click here to view
The area under the ROC curve for NLR was 0.609 and the optimum critical value for NLR was 2.83. AUC for CRP was 0.947 and Optimum critical value of was 6.93 mg/dL.
While the AUC for ferritin was 0.614, the optimal critical value was 182 ng/mL AUC for LDH was 0.554 and the critical value was estimated to be 287.5 U/L. The area under curve for eosinopenia was 0.617 with critical value optimum at 0.95%. The critical value was found to be 6.5 cells/cu mm. D dimer showed promising results with an area under the ROC curve of 0.807. The optimum critical value was calculated as 0.69 pg/mL. The critical values and AUC are shown in [Table 5].
|Table 5: Area under the curve and critical values of individual parameters|
Click here to view
Analysis by receiver operating characteristic for joint indicators
When a combination of NLR and CRP was analysed using the ROC curve, the AUC was 0.728. The optimum critical value was shown to be a predictive probability of 0.41 for combination of NLR and CRP. Area under the ROC curve for a combination of eosinopenia and CRP was 0.732 and critical value was calculated as a predictive probability of 0.44 for a combination of eosinopenia and CRP. The same is shown in [Figure 2] and [Table 6].
| Discussion|| |
We reported here a cohort of 200 patients with laboratory-confirmed COVID-19. Patients were admitted to a tertiary care hospital, Bangalore. We compared the base line lab parameters among cases and controls and tried to deduce the early warning efficiency of these parameters in predicting progression to severe disease.
In a study done by Fan et al. comparison of the hematological parameters between the mild and severe groups showed significant differences in D-dimer, CRP (P < 0.05) and IL-6. These results were consistent with our study with significant difference in the serum levels of D-dimer and CRP among the cases group and controls group in our study. It is a well-established fact that the levels of IL-6 and d dimers are elevated in a severe case of COVID disease and there are several studies substantiating the same. However in terms of early warning efficiency value there is no established evidence that the levels of the same can be of clinical importance in patients who are classified as mild disease and especially IL6 as they are not done routinely at triage for mild patients. However IL6 was not done in many of our cases or controls as our study population was mild at the beginning (cases progressed to severe from mild whereas controls remained mild itself) hence we cannot compare the IL6 levels in our study and the other studies.
A study conducted by a group of authors showed that on admission, older age, lymphopenia and raised LDH were associated with ICU admissions. Patients who were transferred to the ICU had a deeper nadir ALC, nadir hemoglobin, and higher peak ANC and peak LDH levels as compared to patients who did not require ICU stay. The results were comparable to our study where the patients with severe category had higher peak LDH levels.
An ROC curve was plotted for all these lab parameters to see the area under curve of these different lab parameters in having the early warning efficiency in predicting progression of disease and further analysis was made, as ROC curve and the AUC, are an effective measure of accuracy of diagnostic tests and diagnostic ability of biomarkers with proper interpretations Additionally it helps in finding the optimal cut off values, and comparing two alternative diagnostic tasks when each of these is performed on the same patient.
The optimal critical value is that value of the particular investigation which has both the highest sensitivity as well as specificity. The optimal critical value was also determined from the ROC. It was observed in our study that the levels of CRP was high and it was noted that CRP as a marker had an AUC of 0.947, highest among other individual parameters. However, it is of importance to note that the critical value of CRP was 6.93 mg/L. The major drawback of this result is that CRP levels of 5–20 is considered mild and more than 20 mg/L is generally considered high and hence despite CRP having a higher AUC, mild cases with CRP levels between 6.93 mg/L to 20 mg/L will not be seen as a red flag sign in predicting the progression of disease and may give false assurance. In a study conducted by Gao et al., the AUC for CRP was 0.600 which was similar to our study.
The area under the ROC curve for NLR was 0.609 in our study. However in a study conducted by Elshazli et al. AUC was 0.841. In terms of having a good positive predictive value, the results were similar to that of our study. AUC for d-dimer levels were 0.807 in our study. This is comparable to the study done by Gao et al. where the AUC for d-dimers was 0.750.
In our study it was also observed in the cases that lower levels of eosinophils were associated with progression to severity. This interpretation is further strengthened by the following ROC constructed to study the area under curves for Eosinopenia which showed AUC = 0.738.
Since individual markers alone couldn't give the best early warning efficiency a combination of two markers were looked into.
Hence the combination of NLR with CRP was compared with presence of eosinopenia and CRP The latter showed the highest sensitivity and specificity with an AUC of 0.849. indicating better early warning efficiency value of the combination of markers than the markers individually.
ESR, CRP and CBC are among the constituents of routine blood investigations for what is often referred to as a COVID panel. Several hematological and immunological markers, in particular neutrophil count, could be helpful to be included within the routine panel for COVID-19 infection evaluation to ensure risk stratification and effective management. The markers such as eosinophil count (differential and absolute), NLR and CRP are deducible without needing additional investigations such as IL-6 which are not economically and temporally feasible for a large population. IL-6 estimation employs methods of radio-immuno assay and D dimers need reagents to be quantitatively estimated. These act as major drawbacks when estimated for a large population. Estimating the predictive probabilities of joint markers, sensitivity and specificity of the routine investigations have the benefits in terms of cost and time as well.
One of the important limitations in our study was that in majority of our study population IL-6 values were not done at triage unlike the other laboratory markers, since they were all mild cases. Owing to this we couldn't compare it with the different markers as done in other studies. The main strength of this study lies in the fact that the lab values were not compared between patients who directly presented as severe case and mild cases, rather it was between patients two sets who were all mild on admission (Cases progressed to severe from mild and controls remained mild). Thus the lab values done at triage when compared between these two sets is an accurate estimation of early warning efficiency. After literature search it can be noted that not many studies are done with these set of patients in assessing early warning efficiency of the lab parameters This study also emphasizes on using simple markers in unison or in combination, which are easy to perform and economical to predict the set of patients who may worsen and need more vigilant observation from the beginning. This is all the more important when dealing with large number of patients and resources are limited. Further large scale studies can be done and the results can be better interpreted. The results can be incorporated in the many scoring systems being developed in COVID 19 patients to increase its accuracy.
| Conclusions|| |
Different laboratory markers have different early warning efficiency in predicting progression of disease in COVID 19. Among the parameters, CRP, d-dimers, Eosinopenia and NLR have the best early warning efficiency. However, a combination of Eosinopenia and CRP should serve as a red flag sign in patients who may have normal saturation and have minimal symptoms which normally deters us from stratifying them as moderate/severe at triage or when in home isolation. These set of patients need more monitoring and close observation compared to other mild patients.
The authors are grateful to Prof. Dr. N.S. Murthy, Ex-Emeritus Medical Scientist (ICMR), Research Director (M. S. Ramaiah Medical College) for statistical analysis.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Clinical Management Protocol: COVID-19. Government of India Ministry of Health and Family Welfare, Directorate General of Health Services (EMR Division) Version 3, available from www.mohfw.gov.in. [Last accessed on 22 Nov 2020].
Fan BE, Chong VC, Chan SS, Lim GH, Lim KG, Tan GB, et al.
Hematologic parameters in patients with COVID-19 infection. Am J Hematol 2020;95:E131-4.
Yu HH, Qin C, Chen M, Wang W, Tian DS. D-dimer level is associated with the severity of COVID-19. Thromb Res 2020;195:219-25.
Henry BM, de Oliveira MHS, Benoit S, Plebani M, Lippi G. Hematologic, biochemical and immune biomarker abnormalities associated with severe illness and mortality in coronavirus disease 2019 (COVID-19): a meta-analysis. Clin Chem Lab Med 2020;58:1021-8.
Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29-36.
Kumar R, Indrayan A. Receiver operating characteristic (ROC) curve for medical researchers. Indian Pediatr 2011;48:277-87.
Gao Y, Li T, Han M, Li X, Wu D, Xu Y, et al.
Diagnostic utility of clinical laboratory data determinations for patients with the severe COVID-19. J Med Virol 2020;92:791-6.
Elshazli RM, Toraih EA, Elgaml A, El-Mowafy M, El-Mesery M, Amin MN, et al.
Diagnostic and prognostic value of hematological and immunological markers in COVID-19 infection: A meta-analysis of 6320 patients. PLoS One 2020;15:e0238160.
Rodriguez-Morales AJ, Cardona-Ospina JA, Gutiérrez-Ocampo E, Villamizar-Peña R, Holguin-Rivera Y, Escalera-Antezana JP, et al.
Clinical, laboratory and imaging features of COVID-19: A systematic review and meta-analysis. Travel Med Infect Dis 2020;34:101623.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6]