Is glucose-6-phosphatase dehydrogenase deficiency associated with severe outcomes in hospitalized COVID-19 patients?

Glucose-6-phosphate dehydrogenase deficiency (G6PDd) is known to suppress the antioxidant system and is likely to aggravate severity of COVID-19, which results in a pro-oxidant response. This possible association has not been explored adequately in human studies. In this research, we report that the occurrence of non-invasive ventilation, intubation or death—all of which are indicative of severe COVID-19, are not significantly different in hospitalized COVID-19 patients with and without G6PDd (4.6 vs. 6.4%, p = 0.33). The likelihood of developing any of these severe outcomes were slightly lower in patients with G6PDd after accounting for age, nationality, presence of comorbidities and drug interventions (Odds ratio 0.40, 95% confidence intervals 0.142, 1.148). Further investigation that extends to both, hospitalized and non-hospitalized COVID-19 patients, is warranted to study this potential association.

The presenting symptomatology in the study subjects was quite varied and has been summarized in Table 2. Almost two thirds of the patients showed some symptoms at admission in both G6PDd and non-G6PDd groups (59.4% and 63.8% respectively), and cough was found to be the most common symptom in both groups (41.1% and 42.1% respectively). It can be seen that both G6PDd and non-G6PDd patients were comparable with respect to the symptoms on admission.
The drug interventions received by the subjects (Table 3) included Azithromycin, Hydroxychloroquine, Ribavirin, Kaletra, Tocilizumab and convalescent plasma. The number of G6PDd and non-G6PDd patients who received these were similar for most treatments except for azithromycin and hydroxychloroquine. Azithromycin Table 1. Demographic characteristics and comorbidities in the study subjects.

Characteristics
G6PDd present (n = 175) G6PDd absent (n = 1617) p-value Main outcome in the study included requirement of non-invasive ventilation or mechanical ventilation or death. The proportion of patients who suffered either of these outcomes was similar in G6PDd and non-G6PDd patients and ranged between 2.3 and 3.4% as described in Table 4.
The crude odds ratio of suffering the primary outcome for G6PDd patients was found to be 0.69 (95% CI 0.33, 1.45), as shown in Table 5. After adjusting for age, nationality, comorbidities (HTN, CKD, COPD, SCD) and drug interventions, the OR decreased slightly to 0.403 (0.142, 1.148), but remained statistically non-significant (p = 0.132, Supplementary Table S1). We also evaluated whether the co-presence of G6PDd and SCD had differential effect on the outcome, by testing for effect measure modification (Supplementary Table S2). SCD and G6PD did not appear to modify each other's effect on the primary outcome (p = 0.92).

Discussion
The present study did not find hospitalized patients with G6PDd to have higher odds of requiring NIV, HFNC, MIV or death, after accounting for age, nationality, drug interventions and comorbidities (HTN, COPD, CKD SCD).
The patients in both G6PDd and non-G6PDd group were comparable in terms of age and sex. Our cohort was considerably younger with a mean age 45.9 years, compared to hospitalized patients in United States 11,12 (median age 65 years) and United Kingdom 13 (mean age of 70.5 years). The G6PDd groups was predominantly Bahraini nationals as compared to other nationalities, which is not surprising given that the prevalence of G6PDd has been reported to be highly prevalent in the native population of Bahrain from 19 to 26.4% [7][8][9] and to the extent of 47% in those with sickle cell trait 8 . Owing to the same phenomenon, the occurrence of SCD was significantly high in the G6PDd group (9.7%) compared to the non-G6PD group (1.3%). Given recent studies on SCD and risk of severe COVID-19 outcomes 14-16 , we evaluated whether the copresence of SCD and G6PD would affect the outcome by looking at presence of effect measure modification of SCD on the G6PDd and the primary outcome. The result showed that SCD didn't alter the risk of the primary outcome in G6PDd patients (Supplementary Table S2).
The medications used in this cohort should be interpreted with the study timeline. Early in the pandemic HCQ and Azithromycin received wide attention in its use in COVID19 patients 17 . Similarly, Kaletra and Ribavirin in were also used [18][19][20] . Hence, it was common to prescribe these drugs in hospital settings at that time. Moreover, Table 3. Drug interventions during hospital stay.

Medication
G6PDd present (n = 175) G6PDd absent (n = 1617) p-value  www.nature.com/scientificreports/ the use of antibiotics was common in patients with pneumonia. As more evidence about the lack of benefit from these drugs and the potential harm was released, the use of these medications has decreased [21][22][23][24][25][26] . The benefit of steroids in hypoxic COVID19 patients was reported in June 26 and it was only after that that steroids were used in hospitalized patients. HCQ is known to increase risk of acute hemolysis in G6PDd patients and hence it is relatively contraindicated in G6PDd patients. This have led to the reduced use of HCQ in G6PDd patients. On the other hand, Azithromycin was used more commonly in G6PDd patients as the use of HCQ was limited. This has caused more patients in the G6PDd group to receive Azithromycin. The combination between the two drugs have been linked to increased risk of arrythmia, QT prolongation and this could explain the lower use of combination between HCQ and Azithromycin in the non G6PDd patients 27 .
G6PDd group was relatively worse compared to non-G6PDd counterparts with respect to presence of HTN, Asthma and CKD. Meta-analyses have reported HTN to confer a two-fold increase in hazards of mortality (adjusted HR 2.12; 95% CI 1.17, 3.82) 28,29 , while asthma being a respiratory condition is highly associated with worse outcomes in COVID-19 patients with an RR of 1.96; 95% CI 0.89, 4.33 30 . In addition, CKD has been reported to result doubling of the risk of mortality (RR: 2.11; 95% CI 1.72, 2.58) in hospitalized patients 30 . The higher prevalence of these conditions in G6PDd patients kept them at a higher baseline risk to develop worse disease outcomes.
The occurrence of CVD, DM and other Chronic Lung Diseases was not affected by G6PDd status. Diabetes has been established to be a one of the key risk factors severe COVID-19 outcomes and is significantly associated with death or tracheal intubation with 7 days of hospitalization 31 . Prospective studies have reported a two-fold increase in odds of death with CVD (OR 2.46; 95% CI 0.755, 8.044) 28 . Since both these major comorbidities were equally present in both patients with and without G6PD, it is likely that the risk of outcome attributable to these comorbidities would have been apportioned between the groups.
Patients had similar profile and range of presenting symptoms across the G6PDd and non-G6PDd groups with no significant differences. However, it is worth noting that the non-G6PDd group had marginally higher occurrence of symptoms (63.8%) compared to G6PD (59.4%). In addition, relatively larger proportion of the non-G6PDd required mechanical ventilation, non-invasive ventilation as well as oxygenation upon admission and hence, seemed to be worse off compared to the G6PDd group. One of the reasons for this could be that the G6PDd patients could have had a lower admission threshold because of presence of G6PDd and were perceived to be at a higher risk. This could be especially true given G6PDd deficient patients had significantly higher proportion of high-risk comorbidities of SCD, HTN, Asthma and CKD, as well as marginally higher occurrence of CVD.
Initially during the pandemic in Bahrain, patients with risk factors were admitted to hospitals even if they had mild disease. As opposed to cases without risk factors, who were hospitalized based on a clinical decision. Risk factors for severe COVID 19 included: Older Age, Smoking, Obesity, DM, CVD, Respiratory diseases and lung cancers [32][33][34] .
Patients in the G6PDd group were older on baseline and had more of these risk factors, namely CVD HTN and CKD. This has led to a lower hospitalization threshold for the G6PDd patients. This is also reflected upon the severity of disease in both groups. The G6PDd group were less symptomatic and none required advanced oxygenation on admission to the hospital, as most were admitted early during their disease course. Therefore, more patients with G6PDd were hospitalized when compared to non G6PDd patients. Patients who were in home isolation were not studied in this research and hence the outcomes of G6PDd on a population level could not be measured. Owing to these reasons, milder cases of COVID-19 in G6PDd were disproportionately high in this cohort. This has the potential to bias the results towards the null.
One of the main strengths of this study lies in the inclusion of majority of the national tertiary care hospitals that provided care for hospitalized COVID-19 patients. The fact that data collection from the EMR was done manually, reviewed and checked by the study authors and contributors, speaks to the quality of the data. The present study represents the only available research in Bahrain that looks at the role of G6PDd and COVID-19 outcomes.
Our study is restricted to hospitalized COVID-19 patients and therefore it is not possible to generalize the results to COVID-19 cases in the outpatient setting. Due to the same reason, it is not possible to infer the effect of G6PDd on COVID-19 symptomatology in milder cases that did not require hospitalization. Given the data was obtained from EMR, it is possible that the lifestyle factors not captured in EMR such as smoking and obesity, could not be accounted for in the analysis and could be potential confounders. In addition, time from symptom onset, inflammatory markers and frequency of G6PDd related admissions patients were also not captured in the EMR and could be source of unmeasured confounding.
In conclusion, the present study did not find any association between presence of G6PDd and occurrence of serious outcomes in COVID-19 patients. We recommend further large-scale studies that include both hospitalized and non-hospitalized COVID-19 patients, for a more comprehensive insight on the role of G6PDd the pathophysiology of COVID-19 outcomes. www.nature.com/scientificreports/ 18 years of age, had a diagnosis of COVID-19 disease and were admitted between the above specified dates. Data was extracted manually from EMR by five physicians and ten senior medical students, who reviewed all the cases and filled an electronic patient record form developed to collect data for this study. Ethics approval and consent to participate. The protocol and manuscript for this study were reviewed and approved by the National COVID-19 Research Committee in Bahrain (Approval Code: CRT-COVID2020-061). This committee has been jointly established by the Ministry of Health and Bahrain Defense Force Hospital research committees in response to the pandemic, to facilitate and monitor COVID-19 research in Bahrain. All methods and retrospective analysis of data was approved by the National COVID-19 Research and Ethics Committee and carried out in accordance with the local guideline and ethical guidelines of the Declaration of Helsinki 1975. All data used in this study was collected as part of normal medical procedures. Informed consent was waived by the National COVID-19 Research and Ethics Committee for this study due to its retrospective and observational nature and the absence of any patient identifying information.

Data availability
Original data will be made available upon request to the corresponding author.