Antibiotic Resistance related Mortality, Length of Hospital Stay, and Disability-Adjusted Life Years at select Tertiary Hospitals in Uganda: A retrospective study

Background

The Murry et al 2019 review of the global AMR burden showed a glim picture of accelerated progression towards the 2015 predictions on the impact of antimicrobial resistance (AMR) on global health and economic growth (1). Eight years after the prediction that AMR would kill 10 million people annually by 2050, up to 1.27 million deaths in 2019 were attributed to AMR, surpassing the previous leading infectious killers including tuberculosis, malaria and Human Immuno-deficiency Virus (HIV) (1). Over one million of these deaths occurred in the World Health Organization (WHO) African region, of which majority were in Western Sub-Saharan Africa (2). Worryingly, this burden that disproportionately fell on the Africa region, coincided with poor AMR surveillance systems in the region, where the AMR preparedness score was 53% less than the overall Sub-Saharan Africa 2018-2020 Joint External Evaluation (JEE) score, with many countries lacking National Action Plans (NAP) to address the AMR scourge (3). Antimicrobial resistance has dire consequences including the use of more expensive medicines to treat infections and increased length of hospital stay, which lead to higher healthcare costs and reduced economic productivity [9]. Moreover, resistant infections make undertaking of medical procedures like organ transplants, surgeries and chemotherapy risky (4). To reverse this AMR ascendance and its impact, interventions will need to be based on a wider evidence base of the AMR burden including the clinical outcomes, AMR related economic burden, and disability, which is scarce in the WHO Africa region (1).

Despite the growing global attention on AMR, there are substantial limitations in our understanding of the burden, its distribution and the determinants at the population level (5). Currently, majority of the available AMR data describe resistance profiles or mechanism of resistance (6). While this data is important for surveillance and design of treatment protocols, data on AMR burden including clinical outcomes like mortality, disability, length of stay, and their determinant is critical in informing planning, resource mobilization and allocation, policy formulation, and the design of targeted control strategies.

Individuals who survive AMR have increased Disability Adjusted Life Years (DALYs) including the years of life lost (YLL) and year lived with disability (YLD). DALYs measure the burden of disease: reduction in life expectancy and diminished quality of life due to the disease (7). Consequently, studying the distribution of DALYs in the community helps to know which communities bear the greatest burden of the disease and to focus interventions on these areas [10]. Cassini et al found that DALYs due to AMR (170 per 100 000) were similar to DALYs due to influenza, tuberculosis, and HIV combined (183 per 100,000) and 25% (127 of 501 DALYs per 100 000) of the burden of health-care-associated infections (HAIs) was due to AMR (8). Up to 67·9% (115 of 170) of the total DALYs were accounted for by four antibiotic-resistant bacteria: third-generation cephalosporin-resistant Escherichia coli, Methicillin resistant Staphylococcus aureus (MRSA), carbapenem-resistant Pseudomonas aeruginosa, and third-generation cephalosporin-resistant Klebsiella pneumoniae (8). The data indicates a high disease burden posed by AMR.

Determination of the AMR related mortality, length of hospital stay, DALYs and the influencing factors would complement the routine surveillance data that has been accrued during the implementation of the 2018 – 2023 NAP (9–11), and offer a comprehensive description of the AMR burden in Uganda. This study therefore provides a more comprehensive description of the Uganda AMR burden in terms of attributable mortality, DALYs and the driving factors. This information is critical in the campaign to have the Uganda government adopt AMR surveillance in its health prioritization, planning, resource mobilization and allocation.

RESULTS

Data was abstracted from 392 records but 101 records for tuberculosis patients were excluded from this analysis because they heavily skewed the data because of the characteristic long length of stay in hospital. In table 1, socio-demographic characteristics of 291 participants were examined. Among them, 52.6% were males, 10% were living with HIV, and the median age was 35 years (IQR: 21-56 years).

Discussion

We provide the first description of bacterial AMR burden including mortality and DALYs in Uganda, a low-income country. Our evaluation reveals a high mortality: 34.9% due to ARI, of which 30.8% was due to MDRI. Moreover, this mortality occurred in presence of a high rate of comorbidities including HIV (32%) and diabetes (29%) and these patients with comorbidities had a higher mortality (45%) compared to those without (31%). Hospital length of stay was 1.2 times higher for patients with ARI compared to those without. A unit increase in length of hospital stay was associated with a 24% reduction in mortality, indicating early mortality before or shortly after AMR has been diagnosed. Most death were due to sepsis, mainly as a result of Gram-negative bacteria, Escherichia coli and Klebsiella spp in particular as well as Staphylococcus aureus and Enterococcus spp among the Gram-positives, with the ESKAPE group accounting for 85.7% of the deaths. All these contributed to higher YLD and DALYs associated with ARI: 189 and 235.4 compared to 101.7 and 147.0 for non-ARI respectively.

We noted a high AMR related mortality, with a significant proportion being due to MDRI, which is similar to what has been observed elsewhere. Globally, AMR is surpassing previously leading infectious causes of death: 1.27 million deaths and 4.95 million deaths were related or associated to AMR in 2019, higher than that due to malaria, tuberculosis or HIV/AIDS (1). In the same year, in the WHO African region, 1.05 million deaths and 250,000 deaths were due to and attributed to AMR respectively, the highest mortality rate (23.5 per 100,000) globally (2). Resistant Escherichia coli, among the leading six pathogens causing mortality in the same evaluation, has been associated with an absolute risk of mortality of 58 – 130 per 1000 regardless of the type of resistance (17). South Asia shares with Sub-Saharan Africa the brunt of the AMR burden. Overall, case fatality among neonates admitted to the Neonatal Intensive Care Unit (NICU) in Vietnam was 44.4% with a 31.8% 30-day mortality rate, a 27% increase in odds of fetal outcome and a 2.1 day increase in hospital length of stay (18). A systematic review by Kassim et al reported that AMR increased the mortality due to bloodstream infection by 58% and doubled the risk of ICU admissions (19). The risk of death due to AMR affects both community acquired and hospital acquired (CAI, HAI), where the resistant infections in either type have a higher risk of death (20, 21). AMR has been hitherto an unrecognized infectious cause of death in Uganda. In 2019, up to 7100 deaths were due to AMR while another 30700 deaths were associated with AMR, leaving Uganda with the 165^th highest age standardized AMR related mortality among 204 countries (22). Moreover, the AMR related mortality was higher than the previously leading causes of death including tuberculosis, HIV/AIDS and Malaria (22). In this study, the Gram-negative bacteria (Escherichia coli and Klebsiella spp) were responsible for most of the deaths while Staphylococcus aureus and Enterococcus spp contributed most among the Gram-positive. These were among the six leading causes of death in 2019 while others included Streptococcus pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa (1). In the African region, Streptococcus pneumoniae, Klebsiella pneumoniae, Escherichia coli, and Staphylococcus aureus, were the four leading causes of death while third-generation cephalosporin-resistant Klebsiella pneumoniae and methicillin-resistant Staphylococcus aureus (MRSA) were the leading pathogen–drug combinations associated with mortality (2). These findings highlight an urgent need to address the AMR challenge, targeting the pathogens responsible for the highest burden in Uganda and the African region at large.

Diabetes and HIV/AIDS were highly prevalent among AMR patients, and although presence of comorbidities was not independently associated with AMR, it carried a heightened risk of death. Hyperglycemia impairs the immunity (23), increasing the risk for infectious syndromes like skin and soft tissue infections, urinary tract infections, surgical site infection, and respiratory tract infections among others (24). Relatedly, drug resistant infections are also increased, for instance multidrug-resistant Klebsiella pneumoniae and MRSA were more common in hyperglycemic patients compared to normoglycemic patients (25). There has been contrasting evidence on the risk of AMR posed by diabetes: higher AMR was observed in diabetic patients with liver abscess (26) while carbapenem-resistant Klebsiella. pneumoniae was lower among diabetics compared non-diabetics (27). High resistance has been noted in common bacteria isolated from diabetic foot: Staphylococcus aureus against gentamicin (57.96%), and ciprofloxacin (52.45%) and Escherichia coli and Klebsiella Pneumoniae resistance against common antibiotics at more than 50% (28). Overall, AMR is likely to have a higher impact in diabetics due to their heightened risk of infection and frequent exposure to healthcare environment (29). HIV/AIDS was also highly prevalent among patients with AMR in our study. HIV/AIDS weakens the immune system, making individuals prone to infections in addition to the frequent visits to healthcare facilities, which exposes then to infections. HIV/AIDS is associated with high rates of AMR (30): a systematic review showed that people living with HIV (PLWH) were twice as likely to be colonized (31) or infected with MRSA, Streptococcus pneumoniae resistant to penicillin, and third-generation cephalosporins resistant Escherichia coli and Klebsiella pneumoniae (32). Infection prevention measures targeting individuals and hospital or home environment (33), and stewardship practices in the use of antibiotics, in addition to optimized HIV/AIDS care to ensure recovery and sustenance of a robust immune system, are critical measures to protect PLWH from AMR.

On the other hand, we had only few cancer patients in the study as the study was done in a general ward setting. Cancer is a recognized immune suppressive condition including the immune suppressive effect of its treatment while surgery or bone marrow related cancer treatments rely on effective infection prevention or treatment (34). Fifty percent (50%) of cancer deaths are due to infections but deaths due to AMR have not been well described (35). However, high rates of AMR especially in ESKAPE organisms, have been reported in cancer patients: Vancomycin Resistant Enterococcus (VRE, 80%), fluoroquinolone resistant Escherichia coli (70%), and MRSA (50%), among others (36). Consequences of AMR among cancer patients include increased persistence of bacteremia (25% vs 9.7%), metastatic infection (8% vs 4%), and early case-fatality rates (23% vs 11%) among infections due to resistant ESKAPE pathogens versus other bacteria (36). Moreover, MDR is rampant in cancer patients: 40% in ICU patients including 20% Escherichia coli (94.4% ESBL), 12% Staphylococcus aureus (90.6% MRSA), 12% Enterococcus faecium (18.7% VRE), and 6% Acinetobacter baumannii (36). The high burden of comorbidities and associated risk of AMR and the related adverse outcomes highlight the urgent need for strategies to curtail the threat posed by the comorbidities.

Our study’s finding that patients with AMR stayed 1.2 times longer in hospital than patients without AMR, is similar to finding by a study in Ghana, where AMR patients stayed 5 days and 8 days longer than patients with susceptible infections and non-infected patients respectively (19, 37). This is far more than what was found in France, where the excess length of stay due to AMR was only 1.6 days, and 2.9 days in Australia (38, 39). This highlights the difference between the first world and low resource settings. In the first world appropriate diagnostics can aid in early identification of AMR and alternative options of treatment are more available. Long stay in hospital is associated with higher consumption of hospital resources, higher expenditure for the patient, loss of economic productivity and risk of mortality. For instance, in the Ghana study, AMR related costs were USD 1300 higher than those due to susceptible infections, with 30% of the costs being due to loss of productivity through presenteeism and absenteeism (37). Relatedly, the average AMR related cost was less in France (USD 1173) (38) while 40% of AMR related costs in Europe and the United States were also due to loss of productivity (40). A systematic review by Kassim et al estimated a direct cost per AMR case of USD 12,000 in low and middle income countries (LMICs) (19). Hospital acquired infections are more commonly MDR, posing an intense challenge to healthcare. Mauldin et al found that hospital costs and length of hospital stay due to resistant Gram-negative HAI were 29.3% and 23.8% higher than those due to susceptible Gram-negative bacteria (41). On the contrary, however, Wonziak et al did not find any significant difference in length of hospital stay between drug resistant and drug susceptible CAI in Australia (20), as did Barassa et al (42) in Spain. Suzuki et al also found negligible additional length of stay (0.8 days) attributable to resistant Escherichia coli and Klebsiella spp compared to the prolonged length of stay already imposed by bacteremia due to the same organisms (43). In our study every unit increase in time in hospital was associated with a 24% reduction in mortality. Much as the effect of AMR on length of hospital stay has been widely studied, its effect on or relation with mortality has not been fully evaluated. The finding of reduced mortality with a unit increase in length of stay may imply early mortality due AMR before appropriate diagnosis is made or before appropriate treatment is instituted, implying that those who survive through the early events have higher chances of survival.

By accounting for healthy life years lost due to premature death and years lost living with disabilities, DALYs offer a better estimate of burden of disease (44). We found that the DALYs due ARI (235) were almost two times higher than those due to non-ARI (147). AMR related DALYs in our setting were close but higher than DALYs observed in Japan (195) in 2021 (45) but two and four times higher than those in Switzerland (98) and German (57) respectively in 2023 (46), highlighting that low resource settings bear a significantly higher AMR burden. Moreover, the magnitude of DALYs can be affected by type of pathogen (Staphylococcus aureus, Acinetobacter baumannii, and Streptococcus pneumoniae are associated with higher DALYS), sex (higher in male except in women in the reproductive age groups), and site of infection (blood stream infections and bone and joint infection are associated with higher DALYs) while geographical and socio-demographic differences have also been noted (45, 47, 48). These differences likely result from the effect of these parameters on the components used to calculate YLL and YLD particularly number of deaths, incidence of AMR cases, and disability weights. Availability of diagnostic tests and alternative options of treatment for AMR which affect risk of death or the time one stays with the resistant infection, might also explain the difference between high versus low income settings.

We note that the admitting hospital and length of stay in hospital were independently associated with ARI. The three admitting hospitals are located within cities, however, the communities in eastern part of the country are more urbanized than those in the north. Bacterial activity, antibiotic resistant genes, and MDR bacteria in the environment have been noted to increase with urbanization and to differ by level of urbanization, with population size as a key driving factor (49). Conditions in urban centers including; poor housing, poor drainage and poor sanitation; poorly-regulated and private provision of medical services including access to antimicrobials, poor waste management, and increased demand for food-animals associated with inappropriate use of antibiotics as growth-promoters, which drive emergence and transmission drug resistant infection, increase with urbanization (5). Mauldin et al also showed that parameters associated with increased hospital costs and length of stay respectively included pneumonia diagnosis (43.8% vs. 38.2%), ICU admission (142% vs. 106%), neutropenia (83.5% vs. 70.9%) and organ transplant (115.8% vs. 74%) while patients 12 years or more had a 26.3% lower hospital cost but a 66.8% lower length of stay versus those younger (41). Among patients referred for long term care in Chicago, USA, presence of a feeding tube and polymicrobial infection were associated with ARI while urinary catheter, decubitus ulcer, prior use of antibiotics and previous admission within 30 days of current admission were not (50). On the other hand, MDR Pseudomonas aeruginosa infection was independently association with sex, age, and comorbidities, and correlated with ventilation history and neutropenia (51). On the other hand, cotrimoxazole, amikacin, or imipenem except cefazolin; admission to ICU and prior use of acid-suppressive medication were associated with infection with MDR organisms (52). These factors can be targeted by strategies to control AMR and those to improve patient outcomes.

Our study strength lies in it being the first expanded description of AMR burden in Uganda, a resource limited setting, revealing critical data that can inform strategies to control AMR, improve clinical outcomes as well as planning and resource allocation. It was limited in geographical scope as only two regions of Uganda were represented, which may not be representative of other regions in the country. Finally, because of the retrospective design, there was missing data on some of the critical parameters that would have been of interest to evaluate.

Author contribution

Jonathan Mayito, Conceptualization, Investigation, Methodology, Project Administration, Writing – Original Draft Preparation Flavia Dhikusooka, statistical analysis, Review and Editing Daniel Kibombo, Investigation, Methodology, Writing – Review & Editing Alex Andema, Writing – Review & Editing Stephen Obbo, Writing – Review & Editing Alfred Yayi, Writing – Review & Editing Richard Walwema, Resource, Methodology, Supervision, Writing – Review & Editing Francis Kakooza, Resources, Methodology, Supervision, Writing – Review & Editing

Conflict of interest

All authors report no conflict of interest.

Financial disclosure

This work was funded by the Royal Society of Tropical Medicine and Hygiene (RSTMH) Early Career Award in partnership with the National Institute for Health and Care Research (NIHR).