Association Between Treatment with Oral Third-Generation Cephalosporin Antibiotics and Mortality Outcomes in Ebola Virus Disease: A Multinational Retrospective Cohort Study

Abstract

Objective

To evaluate the association between oral third-generation cephalosporin antibiotic treatment and mortality in Ebola Virus Disease (EVD).

Methods

This retrospective cohort studied EVD-infected patients admitted to five Ebola Treatment Units in Sierra Leone and Liberia during 2014–15. Empiric treatment with Cefixime 400 mg once daily for five days was the clinical protocol: however, due to resource variability, only a subset of patients received treatment. Data on sociodemographics, clinical characteristics, malaria status and Ebola viral loads were collected. The primary outcome was mortality compared between cases treated with Cefixime within 48 hours of admission to those not treated within 48 hours. Propensity scores were derived using clinical covariates. Mortality between treated and untreated cases was compared using propensity-matched conditional logistic regression and bootstrapped log-linear regression analyses to calculate an odds ratio (OR) and relative risk (RR), respectively, with associated 95% confidence intervals (CI).

Results

Of 424 cases analyzed, 360 (84.9%) met the Cefixime treatment definition. The mean age was 30.5 years and 40.3% were male. Median Cefixime treatment duration was 4 days (IQR: 3, 5). Among Cefixime-treated patients, mortality was 54.7% (95% CI: 49.6–59.8%), vs. 73.4% (95% CI: 61.5–82.7%) in untreated patients. In conditional logistic regression, mortality likelihood was significantly lower among cases receiving Cefixime (OR=0.48, 95% CI: 0.32–0.71; p=0.01). In the bootstrap analysis, a non-significant risk reduction was found with Cefixime treatment (RR=0.82, 95% CI: 0.64–1.16, p=0.11).

Conclusion

Early oral Cefixime may be associated with reduced mortality in EVD and warrants further investigation.

Introduction

Ebola virus disease (EVD) is a public health threat, with outbreaks increasing in frequency[1, 2]. While there are newly studied specific treatments for Ebola virus (EBOV), which were efficacious during the 2018–19 EVD epidemic in the Democratic Republic of Congo (DRC), there persists high mortality even with the provision of these novel antibody regimes [3, 4] Due to this, EVD guidelines, including those from WHO, recommend a broad suite of supportive care strategies in addition specific EBOV-directed treatment approaches [5].

Empiric treatment with antibiotics is recommended as one such supportive care strategy in clinical guidelines for the management of EVD [3, 5]. The rationale for empiric antibiotic administration in EVD is to mitigate the potential risk of secondary bacterial infections and associated systemic immune dysregulation.[6–9] Recent studies have shown evidence of bacterial translocation across the gastrointestinal wall in EVD, leading to the risk of gram-negative bacteremia [10, 11]. Antibiotics also provide treatment for concomitant bacterial infections common in patients with both suspected and confirmed EVD.[12, 13] Among antibiotics, third-generation cephalosporins are recommended for use in empiric treatment of patients with suspected or confirmed EVD as they act against a broad range of bacterial pathogens, including gram-negative gastrointestinal organisms [5, 14]. Some cephalosporins have anti-EVD properties in vitro.[15] Although scarce, clinical research from the West African outbreak reports frequent use and potential beneficial effects of cephalosporin treatments in patients with EVD [8, 16–19]. These studies, however, were not designed to specifically evaluate the impacts of cephalosporin antibiotics on mortality in patients with EVD, and data to inform clinical care are needed.

During the 2014–2015 West African outbreak, the humanitarian response organization International Medical Corps (IMC) worked with local partners to manage five Ebola Treatment Units (ETUs) in Sierra Leone and Liberia that cumulatively cared for 470 patients with EVD. Demographic, clinical and laboratory data were collected during the course of patient care and combined into a robust multinational database that has been used to evaluate specific types of care provision on patient outcomes in EVD.[20–23] Given the recommended use of empiric antibiotics in EVD care, and the dearth of data on patient-centered outcomes with antibiotic use, this study aimed to evaluate the association between early treatment with oral third-generation cephalosporin antibiotics, in the form of Cefixime, and mortality among patients with EVD admitted to IMC ETUs in West Africa.

Methods

Study Design, Setting, and Population

This retrospective multisite study relied on data collected at five ETUs operated by IMC from 15 September 2014 through 31 December 2015 in Liberia and Sierra Leone.[20] All patients admitted alive to the five ETUs with a final diagnosis of EVD based on real-time reverse transcriptase-polymerase chain reaction (RT-PCR) testing and data on both Cefixime treatment and mortality were eligible for inclusion. The Sierra Leone Ethics and Scientific Review Committee and the institutional review boards of the University of Liberia and Rhode Island Hospital provided ethical approval for this study.

Clinical Procedures

Patients were treated by practitioners trained in standardized ETU care using guidelines developed by IMC in consultation with local and international health authorities (Supplement 1).[24, 25] Empiric treatments included nutritional supplementation, antimalarial medications and broad-spectrum antibiotics plus focused supportive treatments as clinically indicated (Supplement 1). The primary antibiotic treatment regime was an oral third-generation cephalosporin, Cefixime, which had a standing order to be given to all patients admitted to the study ETUs with suspected EVD. The Cefixime protocol was 400 mg once daily for the first five days of ETU care. For children unable to swallow oral medications at baseline, a Cefixime solution was used. Although these guidelines were applied to the greatest extent possible, actual treatments across ETUs varied based on clinical discretion and supply.

In order to be admitted to the EVD-confirmed ward, positive laboratory confirmation by RT-PCR was required. Ebola virus RT-PCR cycle threshold (CT) values (inversely proportional to viral load) were obtained from the United States Naval Medical Research Center Mobile Laboratory for ETUs in Liberia and Public Health England and the Nigerian/European Mobile Laboratory for ETUs in Sierra Leone as previously reported. [26] When completed, malaria testing was performed with the BinaxNow™ rapid diagnostic test (RDT), which assesses for the four main Plasmodium species.

Data Management

Trained providers collected data on demographics and baseline clinical signs and symptoms using standardized forms at the time of triage.[20] Clinical data were recorded one to six times daily on individual patients depending on resource availability. Final disposition was recorded on standardized forms, as described previously.[20, 26] The full cache of data was digitized into a relational electronic database in which fidelity was assessed using Lot Quality Assurance Sampling, which demonstrated 99% consistency with the source records.[20, 27]

The primary outcome was observed mortality during ETU care. Due to the high mortality documented early in ETU care during the West African EVD outbreak [26], and the related potential survivor bias where patients with longer durations of ETU treatment would be more likely to be exposed to different treatments, the primary predictor variable used was early third-generation cephalosporin antibiotic treatment. This was defined as treatment with oral Cefixime 400 mg initiated within 48 hours of ETU admission. Treatment exposure was coded as dichotomous: as received if patients had Cefixime initiated within 48 hours of admission or as not received if they did not meet the treatment definition. Patients with oral Cefixime regimes initiated more than 48 hours post-ETU admission were analyzed as not meeting the pre-defined treatment definition.

Missing data were only present for triage CT values and malaria testing results. Since patients who died were more likely to lack these data, multiple imputations were not appropriate. Instead, to prevent the introduction of bias, CT values were collapsed into three categories: > 22 (low viral load), ≤ 22 (high viral load), and missing [28]. Similarly, malaria RDTs results where coded as positive, negative or missing.

Statistical Analysis

Descriptive analyses were performed for variables using frequencies with percentages, medians with interquartile ranges (IQR), or means with standard deviations (SD) as appropriate. Provision frequency and temporal utilization characteristics of third-generation cephalosporin antibiotics in the population were explored. Bivariable analyses compared demographic and clinical characteristics between groups treated and not treated with Cefixime, and also between patients who died and patients who were discharged alive using Pearson X² or Fisher’s exact tests for categorical variables and by Mann-Whitney or t-tests for continuous variables, as appropriate. Analyses were undertaken using R statistical software version 3.3.3 [29] and statistical significance was set at a value of p<0.05.

Propensity Score Development and Modeling

To control for possible confounding, including the likelihood that patients with coma, vomiting, or dysphagia would be unable to take oral Cefixime, a propensity score model (PSM) was employed to form and compare matched patients treated and not treated with Cefixime during the first 48 hours of care. Variables included in the models were chosen based on their likely relationship to either the treatment exposure or mortality, based on prior research in our dataset and others.[30] The included variables were presence of coma, bleeding, dysphagia, dyspnea, diarrhea, patient age, duration of ETU operation, triage CT value and malaria RDT results. The PSMs used a common support interval to prevent over-extrapolation in which all observations with a probability of receiving treatment less than the minimum in the treated cohort and greater than the maximum in the non-treated cohort were filtered out.[31] Nine patients were removed based on the common support interval parameters.

Cefixime-treated and untreated patients were matched based on the nearest propensity score with replacement, with exact matching on the variables of coma, CT value and malaria RDT results due to the high correlation of those factors with the outcome of mortality.[16, 26, 32, 33] Covariate balance was assessed before and after matching using the calculated standardized bias, to ensure balance was achieved between treatment and control populations. After 1:1 matching, with 56 unique controls, the probability of observed mortality between cases treated and not treated with Cefixime was assessed. Outcomes were compared using a propensity-matched conditional logistic regression model to yield an odds ratio (OR) representing the difference in mortality likelihood between cases treated and not treated with Cefixime and the associated 95% confidence interval (CI). Due to the fact that matching with replacement in conditional logistic regression has the potential to underestimate variance, resulting in a low p-value, a secondary bootstrapped log-linear analysis was performed [34]. Using bootstrapping methods, the complete data set was randomly sampled with replacement using 1000 iterations to calculate a propensity-matched relative risk (RR) estimate for the causal effect of Cefixime treatment on mortality outcomes in the study population.

Results

Characteristics of the Study Population

In the study period there were 478 patients with EVD treated at the 5 IMC ETUs. Of these, 424 met inclusion criteria for analysis (Figure 1). The mean age was 30.5 (SD: ±18.7) years, with a female sex predominance (59.7%). Median ETU care duration was 8 days (IQR: 5, 13). Clinical signs and symptoms present in patients during ETU care are shown in Table 1. The most frequent clinical findings were diarrhea (85.6%), anorexia (80.7%), abdominal pain (76.9%) fever (76.7%) and vomiting (76.7%). Triage CT values were known for 281 patients, of whom 159 (37.5%) had a high viral load (CT value ≤ 22) and 122 (28.8%) had a low viral load (CT value > 22).

Figure 1.

Study population

Table 1.

Characteristics Overall Cohort

	n (%) or mean (±SD)
Age (years)	30.5 (±18.7)
Sex
Female	253 (59.7)
Male	171 (40.3)
Cycle Threshold Results
High Viral Load	159 (37.5)
Low Viral Load	122 (28.8)
Missing Results	143 (33.7)
Malaria Testing Results
Positive	40 (9.5)
Negative	203 (47.9)
Not Tested	181(42.7)
Clinical Characteristics
Abdominal Pain	326 (76.9)
Anorexia	342 (80.7)
Any Bleeding	198 (46.9)
Coma	42 (9.9)
Confusion	57 (13.4)
Diarrhea	363 (85.6)
Dysphagia	249 (58.7)
Dyspnea	205 (48.3)
Fever	325 (76.7)
Vomiting	325 (76.7)

Treatment with Third-Generation Cephalosporin Antibiotics

Of the 424 patients included in the study, 360 (84.9%) patients met the treatment definition of receiving Cefixime within 48 hours of admission. Time of initiation of Cefixime treatments were predominantly early in ETU care as demonstrated by the frequency distribution shown in Figure 2, Panel A. The median treatment duration was 4 days (IQR: 3, 5) with a small proportion of patients receiving Cefixime treatment for more than five days (Figure 2, Panel B). Patients receiving Cefixime during the first 48 hours of ETU care were significantly less likely to have bleeding, coma, and dyspnea during treatment than those not treated (Table 2).

Figure 2.

Characteristics of cefixime Treatments

Table 2.

Patient Characteristics by Cefixime Treatment Within 48 hours of Admission

	No Early Cefixime (n=64) n (%) or mean (±SD)	Early Cefixime (n=360) n (%) or mean (±SD)	p value
Age	28 (±18.3)	31 (±18.8)	0.233
Sex
Female	31 (48.4)	222 (61.7)
Male	33 (51.6)	138 (38.3)	0.055
Cycle Threshold Result
High Viral Load	62 (39.1)	59 (37.2)
Low Viral Load	30.5 (25.0)	36 (29.4)	0.770
Missing Results	51 (35.9)	48 (33.3)
Malaria Testing Result
Positive	8 (12.5)	40 (11.1)
Negative	26 (40.6)	169 (46.9)	0.646
Not Tested	30 (46.9)	151 (41.9)
Clinical Characteristics
Anorexia	43 (67.2)	240 (66.7)	0.936
Any Bleeding	28 (45.3)	92 (25.8)	0.005
Coma	4 (7.8)	1 (0.3)	0.030
Confusion	6 (9.4)	21 (6.1)	0.403
Diarrhea	46 (73.4)	250 (69.7)	0.542
Dysphagia	30 (48.4)	140 (38.9)	0.164
Dyspnea	28 (45.3)	105 (29.4)	0.021
Fever	51 (81.2)	268 (74.7)	0.232
Stomach Pain	35 (54.7)	230 (63.9)	0.177
Vomiting	40 (62.5)	213 (59.4)	0.646

Mortality Outcomes

There were 244 (57.5%) EVD-infected patients who died in the study population. No significant differences in mortality outcomes between the five ETU sites existed, as reported previously.[26] Mortality outcomes differed by viral load at time of ETU admission, where 44.7% of patients who died had CT values ≤ 22 on admission, while this percentage amounted to 27.8% among those who survived. Patients who died were significantly more likely to have diarrhea, dysphagia and dyspnea and showed a trend towards greater bleeding frequency (Table 3). Mortality among patients treated with Cefixime within 48 hours of admission was 54.7% (95% CI: 49.6–59.8%) versus 73.4% (95% CI: 61.5–82.7%) among patients who were not treated (p<0.005).

Table 3.

Patient Characteristics by Mortality Outcome

	Survived (n=180) n (%) or mean (±SD)	Died (n=244) n (%) or mean (±SD)	p value
Age (years)	28.7 (±15.3)	31.8 (+20.8)	0.080
Sex
Female	108 (60.0)	145 (59.4)
Male	72 (40.0)	99 (40.6)	0.906
Cycle Threshold Result
High Viral Load	50 (27.8)	109 (44.7)
Low Viral Load	81 (45.0)	41 (16.8)	<0.001
Missing Results	49 (27.2)	94 (38.5)
Malaria Test Result
Positive	10 (5.5)	30 (12.3)
Negative	101 (56.1)	102 (41.8)	0.005
Not Tested	69 (38.3)	112 (45.9)
Cefixime Treatment within 48 hours of admission
No	17 (9.4)	47 (19.3)	0.005
Yes	163 (90.6)	197 (80.7)
Clinical Characteristics
Abdominal Pain	91 (50.6)	134 (55.3)	0.332
Anorexia	86 (47.8)	119 (48.8)	0.840
Any Bleeding	32 (17.8)	61 (25.4)	0.057
Coma	0 (0)	1 (0.8)	0.158
Confusion	5 (3.3)	13 (5.7)	0.231
Diarrhea	93 (51.7)	152 (62.3)	0.029
Dysphagia	43 (24.4)	90 (36.9)	0.006
Dyspnea	39 (21.7)	80 (32.8)	0.010
Fever	133 (73.9)	183 (75.4)	0.723
Vomiting	86 (48.3)	115 (47.5)	0.872

Propensity-Matched Analysis

Propensity matching achieved covariate balance among all predictors of interest associated with either Cefixime treatment, mortality, or both, allowing for appropriate between group comparative analyses (Figure 3). In the adjusted propensity-matched analysis, mortality prevalence during ETU care was 55.3% among patients treated with Cefixime and 66.1% among patients not receiving treatment within 48 hours of admission. The odds of death were significantly lower in the Cefixime-treated group than the non-treated group based on the conditional logistic regression model (OR=0.48, 95% CI: 0.32–0.71; p=0.01). In the bootstrap analysis, there was a non-significant relative risk reduction of mortality with Cefixime treatment within 48 hours of ETU admission (RR=0.82, 95% CI: 0.64–1.16, p=0.11).

Figure 3.

Love Plot of Standardized Bias Pre and Post Matching

Discussion

In the present data, patients with EVD treated within 48 hours of ETU admission with Cefixime, an oral third-generation cephalosporin antibiotic, had lower mortality than those who did not receive early treatment. The improved mortality outcomes with Cefixime treatment were consistent across the adjusted propensity-matched analyses performed, however in the bootstrap model statistical significance was not achieved. Given these findings, empiric use of antibiotics in patients presenting with EVD in outbreak settings may be beneficial, though further prospective research is required for confirmation.

WHO guidelines recommend provision of broad-spectrum antibiotics empirically to patients being treated for EVD, literature to inform appropriate use of and choice of antibiotic type in EVD care [1, 5] is scarce. Reports from EBOV outbreaks do demonstrate frequent use of antibiotics, commonly with provision of third-generation cephalosporins, which have broad bacterial coverage profiles [8, 16–19]. During the West African outbreak two cohort studies, in which oral or parenteral empiric cephalosporin antibiotics were used in conjunction with a suite of other interventions, did report lower mortality prevalence than was observed in the overall epidemic, however the impacts of the antibiotic treatments were not explicitly evaluated in these studies [16, 19]. The results from the current data are congruent with the literature on use of third-generation cephalosporins and extend the evidence-base by being the first report to demonstrate the potential association between Cefixime treatment and reduced mortality in EVD. As laboratory studies have shown that certain antibiotic compounds, including cephalosporin classes, act against EBOV, the observed beneficial impacts in the studied population are biologically plausible.[15, 35–38] Empiric antibiotic administration may also reduce bacterial translocation from the gastrointestinal tract and thereby prevent secondary infections in patients with EVD [6–11]. Furthermore, as patients admitted to ETUs in outbreak settings have a substantial prevalence of bacterial infections, such as diarrheagenic E. coli and Shigella [13], empiric provision of broad-spectrum antibiotics likely provides appropriate antibacterial treatments to further mitigate mortality risks. The evidence suggests that, in agreement with current guidelines [5], empiric treatment with broad-spectrum antibiotics is appropriate. However, in light of the lack of statistical significance observed in the bootstrap analysis with Cefixime treatment and the lack of patient-oriented data on mortality outcomes with the use of other antibiotic formulations, further prospective studies are warranted.

The potential patient-centered benefits of the provision of broad-spectrum antibiotics in EVD care must be weighed against the public health risk of contributing to advancement of antimicrobial resistance (AMR). AMR is well-recognized problem resulting in excess mortality globally. Approximately 50,000 deaths are estimated to be due to AMR in the United States and Europe annually [39, 40]. Although there are limited data from the Africa region on AMR, the health burdens due to AMR in Africa are likely substantial due to increasing antibiotic usage and insufficient measures to control and monitor appropriate antibiotic access [41–43]. As Gram-negative bacteria have exhibited accelerated resistance profiles [44], and infections with Gram-negative organisms have been commonly identified in ETU populations [13], the theoretical risk of augmenting AMR during outbreak responses does exist. However, as the availability of diagnostic testing for specific infectious diseases in emergency response settings is often poor, and data have demonstrated high mortality among patients admitted to ETUs, including among those testing negative for EVD and/or malaria [33], the benefit of empiric antibiotic regimens likely outweigh the public health risk of AMR in these high-risk patients. Due to the global importance of AMR, though, a crucial consideration in further research on antibiotics in EVD should be evaluation of microbiological resistance profiles among treated patients and associated communities in outbreak settings.

Limitations

There are limitations that should be considered in the current study. The data collection, which occurred during the outbreak response in West Africa, was relatively comprehensive for clinical parameters, however information on laboratory data was missing for some cases. To minimize bias stemming from the missing data, the statistical models were run with cases matched exactly on the status of those variables. Although the propensity models yielded well-matched comparison groups, the models could not take into account all supportive treatments provided across the ETUs, which inhibits the ability to evaluate the impacts of varying care strategies in relation to the Cefixime intervention of interest.

Additionally, the Cefixime exposure groups were not randomized and the specific reasons for non-treatment cannot be identified. Due to this there is potential for residual confounding. However, the propensity-matched approaches were used to reduce confounding by indication, where more critically ill patients are more or less likely to receive treatment.

Due to the small number of cases treated with parenteral third-generation cephalosporin antibiotics, there were insufficient data to accurately evaluate the impacts of non-oral antibiotic treatments in the cohort. Given the substantial gastrointestinal dysfunction and potential for malabsorption of medications that occurs in EVD, this represents an important focus for future clinical research.

Finally, as the data are drawn from patients with EVD treated solely at IMC facilities, the generalizability may be reduced. However, as the practices applied across all IMC ETUs followed international guidelines used by most treatment centers during the West Africa epidemic, these data are likely generalizable to outbreak populations in similarly resource-constrained environments.

Conclusion

The magnitude and high mortality of the West African EBOV outbreak, along with the ongoing 2018–19 outbreaks in the Democratic Republic of Congo and Uganda [1, 2], highlight the imperative for improved understanding of management strategies for EVD. Although novel monoclonal antibody treatments for EBOV now exist [4], supportive care is still a key foundation of clinical management and an important focus for research. The observed reduced mortality with early Cefixime treatment in the multisite and multi-country population studied here suggests that empiric use of oral third-generation cephalosporin antibiotics in patients with EVD may have beneficial patient-centered impacts. However, given the data limitations, additional prospective research is warranted to more comprehensively understand the impacts on patient care of cephalosporin antibiotics, as well as alternative antibiotic classes, and associated evaluation of trends in antimicrobial resistance profiles with use.