Discussion

Our study found that the risk of bacterial infection was higher in patients who had severe COVID-19 at the time of admission to hospital, advanced age, male sex, and presence of chronic kidney disease and malignancy as co-morbid conditions. The most common isolated pathogens were Klebsiella pneumoniae, Acinetobacter baumanii, and Pseudomonas aeruginosa. We found that hospitalized COVID-19 patients with bacterial infections were at higher risk of death and had a longer LOS in addition to other factors. Our study also explored the factors associated with antibiotic use in those who were not identified to have bacterial infections. 60% of these patients received antibiotics and included those who were of advanced age, male sex, having co-morbid conditions. Furthermore, in-hospital mortality was higher, and LOS was more prolonged in these patients. Risk factors for death in these patients included ARDS, lactic acidosis, and thrombotic complications such as myocardial infarction, stroke, and pulmonary embolism.

Several studies have reported higher incidence of bacterial co-infections in severe to critical COVID-19 patients ^{Reference Mirzaei, Goodarzi and Asadi9} and poorer outcomes in these patients. ^{Reference Chen, Zhou and Dong10,Reference Gan, Zhang, Sun and Lyu11} This is postulated to be due to weakened immune system and direct injury to respiratory epithelium by the virus. ^{Reference Cox, Loman, Bogaert and O’Grady12} Risk factors have been variably reported from different parts of the world. A study from China reported greater risk in elderly population and patients with severe disease similar to our results though these variables were not found to be statistically significant on multivariate analysis. They similarly found association with longer length of hospitalization. Most studies have reported an association with severity of COVID-19. ^{Reference Bolker, Coe, Smith, Stevenson, Wang and Reed13,Reference Van Laethem, Pierreux, Wuyts, De Geyter, Allard and Dauby14} A study from Spain reported risk factors for bacterial infection to be male sex, obesity, and multi-organ failure as opposed to our study. ^{Reference López-Herrero, Sánchez-De Prada and Tamayo-Velasco15} Female sex was protective for bacterial infection in a study from Germany which is similar to our finding of increased risk of bacterial infections in male patients. ^{Reference Lösslein, Staus, Beisert Carneiro, Wolkewitz and Häcker16}

Our study adds to the existing body of literature regarding the association of bacterial infections with adverse clinical outcomes including higher mortality and longer hospitalization which has been globally reported from different studies highlighting the need for prompt identification and treatment. ^{Reference López-Herrero, Sánchez-De Prada and Tamayo-Velasco15–Reference Na, Baek and Baek19} A recent large multicenter study from US has reported bacteremic co-infection to be the greatest risk factor for mortality, intensive care admissions, and mechanical ventilation ^{Reference Patton, Orihuela and Harrod20}

Majority of patients were found to have infections with Gram-negative organisms. This is similar to reports from Turkey ^{Reference Iscanli, Aydin and Şaylan17} and Germany ^{Reference Lösslein, Staus, Beisert Carneiro, Wolkewitz and Häcker16} whereas a large study from Spain found Gram-positives to be the most frequent microorganisms notably Streptococcus pneumoniae and Staphylococcus aureus. ^{Reference Moreno-García, Puerta-Alcalde and Letona21} While studies were lacking from LMICs, a study from Indonesia, an upper middle income country reported that the Gram-positive bacteria were predominant pathogens, mainly Staphylococcus aureus, Staphylococcus haemolyticus followed by Gram-negative bacteria Enterobacterales. ^{Reference Ginting, Padmasawitri, Hanum, Nurhayati, Soeroto and Amalia22} In a previous study from our center in the beginning of the pandemic, we had noted increased incidence of infections with Gram-negative organisms ^{Reference Nasir, Rehman and Omair7} which is reflective of our local epidemiology.

Inappropriate antibiotic use during the COVID-19 pandemic has emerged as a major global health concern, with studies showing antibiotic prescribing rates exceeding 70% among hospitalized COVID-19 patients despite bacterial co-infection rates being less than 15%. ^{Reference Rawson, Moore and Zhu23,Reference Langford, So and Raybardhan24} In our study, we found that while 17% were found to have bacterial infection, 60% of the patients hospitalized with COVID-19 who were not found to have a bacterial infection were also administered antibiotics. Studies have reported that this overuse, often driven by diagnostic uncertainty, empirical prescribing practices, and misinformation which were considerably significant issues in LMICs where self-medication and over-the-counter antibiotic access further compounded the problem. ^{Reference Daria and Islam25} Factors associated with antibiotic prescribing in hospitalized COVID-19 patient have been reported from different studies and include advanced age, ICU stay, or need for mechanical ventilation, comorbid conditions such as diabetes mellitus and severe to critical COVID-19 at the time of admission to hospital. ^{Reference Martin, Shulder, Dobrzynski, Quartuccio and Pillinger26–Reference Şencan, Çağ and Karabay28} These are similar to the observations from our study. Interestingly, none of the studies have found antibiotics to be protective of mortality in these patients. ^{Reference Goncalves Mendes Neto, Lo and Wattoo29,Reference Milas, Poncelet, Buttafuoco, Pardo, Lali and Cherifi30} In our study, the mortality was higher, and the length of hospitalization was more prolonged in the patients who got unnecessary antibiotics. This has significant implications for making antibiotic stewardship guidelines and for future pandemic preparation particularly in resource constrained settings.

There is a paucity of data from LMICs regarding antibiotic use during the pandemic, whereas most existing literature is skewed toward high-income settings. Our study helps to fill this knowledge gap and supports evidence-based global health recommendations that are more inclusive and representative. However, there are limitations such as generalizability as our study is a single-center study, and the findings may be applicable to tertiary care centers in LMICs. Furthermore, since this is retrospective data, there may be residual confounding from factors that could not be captured such as differentiating colonization from infection. However, we mitigated that by using a sensitive case definition to not miss anyone with infection. To address any potential misclassification, we also conducted a sensitivity analysis restricting bacterial infection to cases with microbiological confirmation only and excluding those identified by PCT > 5 ng/mL alone. Our findings remained consistent with our initial analysis. In our study, we defined severe illness as requiring noninvasive or mechanical ventilation as an objective marker of significant respiratory compromise. Although narrower than WHO criteria, which include oxygen thresholds, this definition offers a pragmatic approach given the available data and is a clinically meaningful proxy for severe illness. Hence, our findings are most directly applicable to patients requiring advanced respiratory support. Our study did not have detailed data on antibiotic duration (beyond initial prescribing). However, our institution has a well-established Antimicrobial Stewardship Program (ASP) to limit inappropriate or extended antibiotic courses.

Our study findings can be used to tailor antimicrobial stewardship (AMS) programs in tertiary care centers in LMICs as they reflect local prescribing behavior and can inform interventions that will be practical, sustainable, and effective in LMIC settings.