The interaction effect between hemoglobin and hypoxemia on COVID-19 mortality in a sample from Bogota, Colombia: An exploratory study.

Purpose: We aimed to assess the effect of hemoglobin (Hb) concentration and oxygenation index on COVID-19 patients' mortality risk.Patients and methods: We retrospectively reviewed sociodemographic and clinical characteristics, laboratory findings, and clinical outcomes from patients admitted to a tertiary care hospital in Bogota, Colombia. We assessed exploratory associations between oxygenation index and Hb concentration at admission and clinical outcomes. We used a generalized additive model (GAM) to evaluate the nonlinear relations observed and the classification and regression trees (CART) algorithm to assess the interaction effects found.Results: From March to July 2020, 643 patients were admitted, of which 52% were male. The median age was 60 years old, and the most frequent comorbidity was hypertension (35.76%). The median value of SpO2/FiO2 was 419, and the median Hb concentration was 14.8 g/dL. The mortality was 19.1% (123 patients). Age, sex, and history of hypertension were independently associated with mortality. We described a nonlinear relationship between SpO2/FiO2, Hb concentration and neutrophil-to-lymphocyte ratio with mortality and an interaction effect between SpO2/FiO2 and Hb concentration. Patients with a similar oxygenation index had different mortality likelihoods based upon their Hb at admission. CART showed that patients with SpO2/FiO2 < 324, who were older than 62 years, and had an Hb of >= 16 g/dl had the highest mortality risk (96%). Additionally, patients with SpO2/FiO2 > 324 but Hb of < 12 and neutrophil-to-lymphocyte ratio of > 4 had a higher mortality likelihood (57%). In contrast, patients with SpO2/FiO2 > 324 and Hb of > 12 g/dl had the lowest mortality risk (10%).Conclusion: We found that a decreased SpO2/FiO2 increased mortality risk. Extreme values of Hb, either low or high, showed an increase in likelihood of mortality. However, Hb concentration modified the SpO2/FiO2 effect on mortality; the likelihood of death in patients with low SpO2/FiO2 increased as Hb increased.


56
On December 31, 2019, SARS-CoV-2 emerged in China and rapidly spread worldwide, with more 57 than 293 million infections and 5.4 million deaths.
[1] In Colombia, which is a developing country, 58 as of January 5th, there have been more than 5.2 million cases, with more than 130,000 deaths 59 in the national territory. In Bogota, the most affected city in the country, there have been more between high altitude and COVID-19 mortality and lower mortality excess. [11][12][13] People who 76 permanently live in high-altitude places develop adaptative mechanisms against low atmospheric 77 oxygen pressure exposure. In locations such as Bogotá, which has an altitude of more than 2,500 78 meters, there is a decrease in mean SpO 2 values among its population. [14] People who live in . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 9, 2022. ; https://doi. org/10.1101org/10. /2022 106 interrelated moderating the disease outcome. Our research aimed to assess the effect of Hb and 107 oxygenation index on COVID-19 mortality.

Materials and methods
109 Study population 110 We performed an observational, retrospective study between March and July 2020. We followed 111 a cohort of COVID-19 patients admitted to the Hospital Universitario Mayor Méderi (HUM) in 112 Bogotá, Colombia. In this study, we used a cutoff date of July 2020. We included patients with 113 confirmed infection of SARS-CoV-2 by RT-PCR or antigen. We excluded patients with a history 114 of anemic, lymphoproliferative, or myelodysplastic syndromes. We also excluded patients with . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 9, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 9, 2022. terms. Moreover, we found a significant interaction term between SpO2/FiO2 and hemoglobin.
190 Table 2 shows the effects of the linear terms in the independent and interaction GAM. These 191 estimations were the same for both models. Table 3 shows the smoothed variables in both 192 models and the significance of its smoothed function.

193
194 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 9, 2022. ; https://doi.org/10. 1101/2022

203
It is noteworthy that the model with the interaction term had a higher Akaike information criterion 204 (AIC) and a lower determination coefficient (R 2 ) than the principal effects model, which came at is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 9, 2022. Figure 1 shows the functional form of the partial effects on mortality of the smoothed terms in

222
The contour plot in Figure 2 shows how Hb modifies the SpO 2 /FiO 2 effect on mortality.

240
Notice how decreases in SpO2/FiO2 get worse as hemoglobin concentration rise. Blanks in the 241 plot represent data that were not observed.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 9, 2022. ; https://doi.org/10.1101/2022.02.07.22270640 doi: medRxiv preprint 242 243 Figure 3 shows a CART model, a method that classifies patients according to their likelihood of 244 death using SpO 2 /FiO 2 , Hb, age, and NLR as independent variables. The patients with the 245 highest probability of death (90%) were those who had SpO 2 /FiO 2 <324, were older than 62 246 years, and had an Hb >16 g/dL. Additionally, patients with an Hb <12 g/dl and NLR >4 had a 247 higher probability of death (57%). In comparison, patients with the lowest likelihood of death 248 (10%) were those who did have an SpO 2 /FiO 2 >324 and an Hb >12 g/dL at admission.

249
Interestingly, the CART method also showed that the hemoglobin concentration at admittance 250 modified the SpO 2 /FiO 2 effect on probability of death. sharp quadratic increase in the likelihood of death (Fig. 1). We found a significant interaction . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

274
The proportion of death among hospitalized patients was 19.12%, and the cumulative death

288
We found a significant association between SpO 2 /FiO 2 and mortality. The GAM showed a sharp 289 increase in the probability of mortality as SpO 2 /FiO 2 decreased (Fig. 1) based on oxygenation compromise. There is a correlation between disease severity measured by . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 9, 2022.

305
Red bloodline characteristics have raised concern in the physiopathology of COVID-19 and

306
ARDS. In our study, either high or low Hb values were associated with higher likelihood of death,

307
giving a nonlinear, U-shaped correlation between hemoglobin concentration and probability of 308 death in GAM (Fig. 1C). However, the oxygenation index measured by the SpO 2 /FiO 2 ratio 309 modified this effect. Patients with similar SpO 2 /FiO 2 had different mortality probabilities according 310 to Hb concentration (Fig. 2). against iron. This leads to a functional loss of hemoglobin and hemolysis. However, given that 319 these hypotheses proposed a nondescribed protein-protein hemoglobin degradation pathway, 320 some authors criticized it for its flawed methods and unreliability. [33,34] In the same way, De . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 9, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 9, 2022. intervention requirements that we did not observe, and which could modify the prognosis.

383
Additionally, we did not consider certain variables related to the Hb dissociation curve as PCO 2 or 384 acid-base status, which might be associated with residual confusion bias. Furthermore, we did 385 not consider the pharmacological history that could affect the baseline CBC. Lastly, given the 386 retrospective nature of our study, and since the data were gathered from clinical records, we 387 could not guarantee a uniform measurement for each variable. It is necessary to note that no 388 effect can be associated with high-altitude exposure given the absence of a comparative group.

389
The variables studied might also be modified in chronic pulmonary disease or tobacco use. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

409
Author contributions 410 AMPR and AMRS led and supervised the execution of the research. They contributed to 411 evidence interpretation, critical reviewing and commenting on the manuscript.

412
DRRL conceptualized the study, conducted the project, collected the data, interpreted the 413 evidence, and critically reviewed the manuscript.

414
NMG contributed to data curation, formal analysis, and visualization.

415
AFPA conceptualized the study, worked in data curation, formal analysis, visualization, evidence 416 interpretation, and manuscript drafting. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 9, 2022. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 9, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 9, 2022. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 9, 2022. ; https://doi.org/10.1101/2022.02.07.22270640 doi: medRxiv preprint