Skip to main content
Download PDF

Antibiotics versus Non-Antibiotic in the treatment of Aspiration Pneumonia: analysis of the MIMIC-IV database

BMC Pulmonary Medicine volume 24, Article number: 621 (2024) Cite this article

Abstract

Background

Aspiration pneumonia (AP) is a common complication in the intensive care unit (ICU), which is associated with significantly increased morbidity and mortality and has a significant impact on patient prognosis. Antibiotics are commonly used in the clinical treatment of AP. However, the prognostic impact of antibiotics on patients with AP has not been adequately characterized. The purpose of this study is to illustrate the relationship between the use of antibiotics and in-hospital mortality of AP patients, as well as to analyze the effects of different antibiotic treatment regimens on the prognosis of the patients, and to further understand the distribution of pathogens and drug resistance in AP patients, so as to provide guidance information for the rational use of medication for patients in the clinic.

Methods

Clinical data of AP patients were extracted from the MIMIC-IV database. Statistical methods included multivariate logistic regression, propensity score matching (PSM), and inverse probability weighting (IPW) based on propensity scores to ensure the robustness of the findings. In addition, the characteristics of the medications used by patients with AP were described using statistical graphs and tables.

Results

A total of 4132 patients with AP were included. In-hospital mortality was significantly lower in the group using antibiotics compared to the group not using antibiotics (odds ratio [OR] = 0.44, 95% confidence interval [CI] 0.27- 0.71, P = 0.001). Furthermore, in the group using mechanical ventilation (MV), antibiotics use significantly reduced in-hospital mortality (OR = 0.30, 95% CI 0.15–0.57, P < 0.001).

Vancomycin and cephalosporins are the most commonly used antibiotics to treat AP. Specifically, vancomycin in combination with piperacillin-tazobactam was used most frequently with 396 cases. The highest survival rate (97.6\%) was observed in patients treated with levofloxacin combined with metronidazole. Additionally, vancomycin combined with piperacillin-tazobactam had many inflammation related features that differed significantly from those in patients who did not receive medication.

Conclusions

Antibiotics use is closely associated with lower in-hospital mortality in ICU patients with AP. Moreover, understanding antibiotics use, the composition of pathogenic bacteria, and the rates of drug resistance in patients with AP can aid in disease prevention and prompt infection control.

Peer Review reports

Background

Aspiration pneumonia (AP) is a common hospital-acquired infection. It is a frequent complication in stroke patients and among those admitted to the Intensive Care Unit (ICU). The pathophysiology of AP primarily involves dysphagia and reflux of gastric contents. Once AP occurs, it can further exacerbate the disease process and have a severe impact on the patient’s prognosis. The incidence of AP in the ICU is reported to vary from 5.0% to as high as 67%. The mortality rate of patients with AP exceeds 14% [1]. Data from Korea show that the 1-year, 3-year, and 5-year mortality rates for AP patients are 49.0%, 67.1%, and 76.9%, respectively [2]. Patients with AP often exhibit increased disease complexity and an extended need for intensive care. This increases the economic burden on both the patient and the healthcare system [3]. Fur- thermore, in clinical treatment, antibiotics are the primary medication for treating patients with AP [4]. Given the correlation between the pathogens of AP and the nor- mal flora of the gastrointestinal tract, this association should be considered when selecting antibiotics [5, 6]. However, medical personnel often formulate antimicrobial treatment plans for patients based on personal experience. To reduce the increased incidence of disease and consumption of medical resources caused by subjective bias, and to enhance the rationality, safety, and effectiveness of clinical medication, and to quickly transform empirical treatment into targeted therapy, it is crucial for healthcare professionals to develop more objective pharmacological treatment plans for patients with AP [7].

This study utilizes clinical characteristics, the distribution of pathogens in respi- ratory secretions and their resistance properties, the change in inflammatory response indicators, to provide a reference for clinical medication guidance for patients with inhalation pneumonia in the ICU.

Methods

Data source

The Medical Information Mart for Intensive Care (MIMIC) database was established in 2003 through collaboration and funding from multiple institutions, including Beth Israel Deaconess Medical Center (BIDMC), and the National Institutes of Health Technology, Massachusetts General Hospital, emergency room physicians, critical care physicians, computer science experts, and other professionals in the field of critical care medicine database [8]. MIMIC is recognized as the largest open-source and freely accessible clinical database in the fields of critical care and emergency medicine. It is based on the intensive care inpatient system at Beth Israel Deaconess Medical Center. This database effectively addresses the lack of extensive systematic clinical diagnostic and treatment data that clinical practitioners face in research. The latest version, MIMIC-IV (version 2.2), contains data from 2008 to 2019 [9,10,11]. In this study, the Collaborative Institutional Training Initiative (CITI) course was thoroughly com- pleted, and authorization to access the MIMIC-IV database was obtained (Record ID: 52,570,571). Since this project does not affect clinical treatment, and all protected health information has been anonymized, the requirement for individual patient consent was waived.

Study population

Patients diagnosed with AP were identified in the MIMIC-IV database using ICD-9 codes (5070, 5071, and 99,732) and ICD-10 codes (J690, J691, J698), and the records of all adult patients over 18 years old admitted to the ICU with AP were analyzed. For patients admitted to the ICU multiple times, only the first admission was selected, and those who died within 24 h of admission were excluded. It should be noted that these patients with AP include those diagnosed with community-acquired and hospital-acquired AP.

Data extraction

The collected data include demographic characteristics, comorbidities, treatment measures, vital signs, laboratory measurements, severity of illness, duration of first ICU admission, length of hospital stay, in-hospital mortality, microbiology, and antibiotic usage information. Data were retrieved from MIMIC-IV using Structured Query Lan- guage (SQL) and Navicat15. Treatment measures refer to their corresponding usage within 24 h after ICU admission. Vital signs and laboratory measurements refer to the first measurements taken from 6 h before ICU admission to 24 h after admission. The SOFA score is calculated within the timeframe from 6 h before ICU admission to 24 h after admission. Antibiotic use refers to the administration of antibiotics from 24 h before ICU admission to 48 h after admission.

It was observed from the selected data that more than half of the patients did not have recorded values for heart rate and albumin. If heart rate and albumin values were used directly as variables, a large number of missing values would be present. Therefore, whether or not heart rate and albumin values are recorded will be used as variables [12].

Statistical analysis

Univariate analyses were conducted for all study variables. Continuous variables were expressed as median and interquartile spacing (IQR) and compared using the Wilcoxon rank sum test. Categorical variables were represented by counts and percentages (%) and were compared using the chi-square test. To evaluate the effectiveness of the Propensity Score Matching model (PSM) in balancing the distribution of variables between the antibiotic use group and the non-antibiotic use group, the Standardized Mean Difference (SMD) between these groups was calculated. Multiple imputation was performed on variables with a missing rate less than 50% using the’mice’ package in R [13, 14]. An additional table file shows which variables were missing (see Additional file 1: Table S1).

To ensure the robustness of this study results, PSM and Inverse Probability Weight- ing based on propensity scores (IPW) were used [15,16,17,18]. In the PSM model, a gradient boosting model (GBM) was used to estimate the propensity scores of patients (reflect- ing the probability of an individual being assigned to the treatment or control group), minimizing the imbalance invariable distribution between the antibiotic use group and the non-use group. GBM is a machine learning algorithm that progressively enhances model performance by combining multiple base learners. The main idea is to train new learners in each iteration based on the errors of the previous round, then combining them into a powerful overall model [19,20,21]. In our study, regression trees were used as the base learners for the GBM model, which included a total of 35 covariates: (1) Demographic characteristics [age (years), gender, ethnicity]; (2) Comorbidities (con- gestive heart failure, cerebrovascular disease, myocardial infarction, dementia, chronic pulmonary disease, malignant tumor, Charlson Comorbidity Index); (3) Treatment measures (mechanical ventilation, vasopressor drugs); (4) Vital signs (heart rate, res- piratory rate, body mass index); (5) Laboratory measurements (hemoglobin, platelets, white blood cells, red blood cells, arterial oxygen pressure, arterial carbon dioxide pressure, blood urea nitrogen, creatinine, glucose, potassium, albumin, INR value, lymphocytes, monocytes, neutrophils, pH value); (6) Severity of illness (SOFA score, APSIII score); (7) Duration of first ICU admission; (8) Length of hospital stay. An additional figure file shows the contribution of each covariate to the final propensity score (see Additional file 2: Figure S1). In the PSM model, one-to-one nearest neigh- bor matching was employed with a caliper width of 0.00055. The results are presented as Odds Ratios (OR) and their corresponding 95% Confidence Intervals (95% CI).

For the IPW model, observational data are adjusted through propensity scores to create a weighted pseudo-population sample. The purpose of this approach is to reduce bias introduced by unbalanced samples, more accurately estimate treatment effects or control for confounding factors, thereby enhancing the internal validity and reliability of our study. The results are presented as OR and their corresponding 95% CI.

In the PSM cohort, multivariate logistic regression analysis was conducted to eval- uate the association between antibiotic use and in-hospital mortality, with results presented as OR and their corresponding 95% CI.

In the PSM cohort, subgroup analyses was conducted to explore whether the asso- ciation between antibiotic use and in-hospital mortality is influenced by age, gender, race, severity of illness (SOFA score, APSIII score), treatment measures (mechanical ventilation), and respiratory rate.

In the PSM cohort, Kaplan–Meier (K-M) survival analysis method was conducted to plot the survival curves for the antibiotic use group and the non-antibiotic use group within 30 days after hospital admission, to evaluate the impact of antibiotic use on the survival of patients with AP, thereby further determining the effectiveness of antibiotics in treating patients with AP [22].

Statistical graphs (including bar graphs, line graphs, and Venn diagrams) were used to represent the distribution of data, and statistical tables were used to describe differences between patients. A two-tailed test with a p − value < 0.05 was considered statistically significant.

Results

Population and baseline characteristics

A total of 5271 patients with AP in ICU were enrolled in this study and according to the exclusion criteria, 4132 eligible patients in ICU were selected. Of these, 3429 patients (82.99%) used antibiotics from 24 h before ICU admission to 48 h after admission, while 703 patients (17.01%) did not use antibiotics during this period. After adjustment using PSM, 256 patients who used antibiotics and 256 patients who did not use antibiotics were included in the PSM cohort for analysis (Fig. 1). The characteristics and outcomes of the original cohort and the PSM cohort are shown in Table 1. Patients who used antibiotics had significantly higher severity of illness scores: SOFA score was 5.00 [3.00, 8.00] vs. 4.00 [2.00, 6.00], and APSIII score was 49.00 [37.00, 63.00] vs. 40.00 [29.00, 53.00]. There were significant differences between the antibiotic use group and the non-antibiotic use group in laboratory measurements, except for glucose levels. Within 24 h of ICU admission, a larger proportion of patients with AP received mechanical ventilation (68.8% vs. 59.3%). A higher proportion of patients with AP had congestive heart failure (32.6% vs. 26.2%), cerebrovascular disease (18.8% vs. 36.3%), and chronic pulmonary disease (29.6% vs. 23%).

Fig. 1
figure 1

Flowchart of patient selection for the study. MIMIC-IV: Medical Information Mart for Intensive Care Database IV; ICU: intensive care unit

Full size image
Table 1 Baseline characteristics between groups before and after PSM
Full size table

Relationship between antibiotic use and in-hospital mortality, 30-day post-admission survival rate

As shown in Table 1, the overall in-hospital mortality rate in the original cohort was 20.5%, with the antibiotic group experiencing a mortality rate of 20.9%, and the non-antibiotic group a rate of 18.3%. After the implementation of PSM, the effect of confounders was successfully minimized, ensuring higher comparability between the study and control groups. In the cohort adjusted for confounders through PSM, the overall in-hospital mortality rate was 16.4%, with the in-hospital mortality rate in the antibiotic group at 10.9% and in the non-antibiotic group at 21.9%.

As shown in Fig. 2, after adjusting for confounding factors through PSM, the OR for antibiotic use in the multivariate logistic regression was 0.37 (95% CI 0.20–0.67, P = 0.001). For IPW, the OR was 0.80 (95% CI 0.70–0.90, P < 0.001), and for PSM, the OR was 0.44 (95% CI 0.27–0.71, P = 0.001).

Fig. 2
figure 2

Association between antibiotic use and in-hospital morality of ICU patients. OR: odds ratio; CI: confidence interval; ICU: intensive care unit; Multivariate: adjusted for all the baseline variables shown in Table 1. PSM: propensity score matching. IPW: inverse probability of treatment weighting

Full size image

As illustrated in Fig. 3, after adjusting for confounding factors using PSM, a 30- day post-admission survival analysis was conducted for both the antibiotic use and non-use groups using the K-M method. The survival rate of the antibiotic use group was significantly higher than that of the non-use group, with a Log-rank P < 0.001.

Fig. 3
figure 3

Kaplan–Meier indicates the association between antibiotic use and 30-day survival of patients presented with Aspiration Pneumonia

Full size image

The aforementioned analysis indicates that the use of antibiotics significantly ben- efits both reducing in-hospital mortality and improving the survival rate within 30 days post-admission.

Subgroup analysis

The number of patients in each subgroup is shown in Fig. 4. Except for the groups with SOFA > = 5, APSIII > = 69, and those not using mechanical ventilation, the use of antibiotics significantly reduced in-hospital mortality in patients with AP in all other subgroups. Compared to patients aged > = 65 years (OR 0.53, 95% CI 0.29- 0.94, P = 0.033), patients aged < 65 years (OR 0.31, 95% CI 0.11–0.74, P = 0.012) had a lower OR value.

Fig. 4
figure 4

Association between antibiotic use and in-hospital morality in subgroup. OR: odds ratio, CI: confdence interval

Full size image

Antibiotic usage in patients with AP

Figure 5 illustrates the statistical overview of antibiotic usage in patients with AP. The most frequently used antibiotic is vancomycin, followed by fourth-generation cephalosporins (ForGC). Antibiotics used by fewer than 80 patients are categorized as’Other’ due to their relatively small number in comparison to the total.

Fig. 5
figure 5

Antibiotic used in patients with Aspiration Pneumonia. FirGC: First-Generation Cephalosporins, ThirGC: Third-Generation Cephalosporins, ForGC: Fourth-Generation Cephalosporins

Full size image

Figure 6 depicts the combined use of antibiotics and the corresponding patient num- bers. Among them, the largest number of patients received a combination of two antibiotics (N = 1154), followed by patients treated with a combination of three antibi- otics (N = 1023), while the least number of patients were treated with a combination of eight antibiotics (N = 1).

Fig. 6
figure 6

Patterns in Antibiotic Usage and Patient Counts. The columns show the number of antibiotics used, and the broken lines shows the number of people taking drugs. The horizontal axis shows the number of antibiotics used. 0 indicates that no antibacterial drugs have been used, one indicates that only one antibacterial drug has been used, and two indicates the combination of two antibacterial drugs, and so on for 3, 4, 5, 6, 7,and 8

Full size image

Figure 7 shows the most commonly used five antibiotics (vancomycin, ForGC, piperacillin-tazobactam, metronidazole, third-generation cephalosporins (ThirGC)) and their combinations in patients with AP. The most common combination ther- apy was vancomycin + piperacillin-tazobactam (N = 396), followed by vancomycin + ForGC (N = 246). Only 3 patients received a combination of all five antibiotics. 102 patients were treated solely with ThirGC, and a relatively smaller number of patients were treated exclusively with metronidazole.

Fig. 7
figure 7

Distribution of different antibiotics and their combinations in patients with Aspiration Pneumon

Full size image

As illustrated in Fig. 8, we further analyzed the usage, patient numbers, and sur- vival rates associated with antibiotics. The results indicate that the combination of vancomycin and piperacillin-tazobactam was the most commonly used, while the combination of levofloxacin and metronidazole had the highest survival rate at 97.6%.

Fig. 8
figure 8

Antibiotic Utilization, Patient Numbers, and Survival Rates. The columns show the number of people who took drugs. The broken line shows the survival rate. None indicates no antibiotic usage, VAN represents Vancomycin, 4GC represents Fourth-Generation Cephalosporins, TZP represents Piperacillin-tazobactam, MNZ represents Metronidazole, 3GC represents Third-Generation Cephalosporins, LVX represents Levofloxacin, AZM represents Azithromycin, 1GC represents First-Generation Cephalosporins, and SAM represents Ampicillin-sulbactam

Full size image

To investigate the reasons for medical staff using antibiotics, we compared the differences between vancomycin + piperacillin-tazobactam (Table 2) or levofloxacin + metronidazole (Table 3) and no antibiotics. This revealed that only one inflammation indicators of patients who used levofloxacin + metronidazole differed from those in the no antibiotics group, and that all inflammatory indicators of patients who used vancomycin + piperacillin-tazobactam differed from those in the no-antibiotics group.

Table 2 Comparison of differences in inflammatory related characteristics: piperacillin-tazobactam + vancomycin vs. Unused antibiotics
Full size table
Table 3 Comparison of differences in inflammatory related characteristics: Metronidazole + Levofloxacin vs. Unused
Full size table

Composition of pathogenic bacteria and drug resistance rates in patients with AP

Table 4 lists the distribution of pathogens in respiratory specimens of patients with AP treated with vancomycin + piperacillin-tazobactam. In the 396 AP patients treated with vancomycin + piperacillin-tazobactam, 2051 strains of pathogens were detected. These included 156 strains (7.60% of the total) of fungi, 660 strains (32.18% of the total) of Gram-positive bacteria, and 1235 strains (60.22% of the total) of Gram- negative bacteria.

Table 4 Aspiration Pneumonia: Main Pathogenic Bacterial Distribution with Vancomycin and Piperacillin-Tazobac
Full size table

Tables 4 and 5 show the drug resistance characteristics of Gram-positive bacteria in respiratory specimens of patients with AP treated with vancomycin + piperacillin-tazobactam. The results indicate that out of 660 strains of Gram- positive bacteria, 192 strains (29.1% of the total) were resistant, including 152 strains of coagulase-positive methicillin-resistant Staphylococcus aureus, 37 strains of methicillin-resistant Staphylococcus aureus, and 3 strains of resistant Streptococcus pneumoniae.

Table 5 Aspiration Pneumonia: Resistance of Primary Gram-Positive Bacteria to Vancomycin Combined with Piperacillin-Tazobactam
Full size table

An additional table file shows the drug resistance characteristics of Gram-negative bacteria in respiratory specimens of patients with AP treated with vancomycin + piperacillin-tazobactam (see Additional file 3: Table S2). The results indicate that out of 1235 strains of Gram-negative bacteria, 169 strains were resistant (13.68% of the total), including 20 strains resistant to piperacillin-tazobactamorvancomycin, though the resistance rate was relatively low (< 1.62%).

An additional table file lists the distribution of pathogens in respiratory specimens of patients with AP treated with levofloxacin + metronidazole (see Additional file 4: Table S3). In the 41 patients with AP treated with levofloxacin + metronidazole, 41 strains of pathogens were detected. These included 4 strains (9.76% of the total) of fungi, 35 strains (85.36% of the total) of Gram-positive bacteria, and 2 strains (4.88% of the total) of Gram-negative bacteria. An additional table file outlines the drug resistance characteristics of these pathogens. In the respiratory specimens of patients with AP treated with levofloxacin + metronidazole, 2 strains were resistant to levofloxacin, and 2 strains were MRSA positive (see Additional file 5: Table S4).

Discussion

This is the first retrospective study in the field of critical care and emergency medicine based on the clinical characteristics and medication features of patients with AP, using the largest and most recent open database (MIMIC-IV database). It aims to investigate the effect of antibiotic use on mortality in patients with AP, as well as the prevalence of antibiotic use and bacterial infections in these patients. The goal is to provide medical evidence for pharmacological interventions. Mortality rates were compared between patients who used antibiotics and those who did not. The impact of different antibiotic regimens on patient prognosis and inflammation-related markers was analyzed, and further insights were gained into the distribution of pathogens and drug resistance among patients with AP.

The application of antibiotics has had a positive impact on the mortality rate of AP in the Intensive Care Unit (ICU), making it a treatable disease in most cases. Antibiotics can also shorten the duration of the disease and hospital stay. This study shows that the use of antibiotics is significantly associated with a reduction in the in- hospital mortality rate of patients with AP in the intensive care unit. Similar to our study, current research confirms that antibiotics may play a positive role in reducing the mortality rate of critically ill patients. First, a recent study found that the use of antibiotics can reduce the mortality rate of patients with AP [23, 24]. Recent research suggests that antibiotics can stop the progression of infection and promote the recov- ery of critically ill patients by reducing bacterial load and inflammation and restoring lung function [25]. Moreover, antibiotics have been found to modulate the host immune response, reduce the severity of lung infections, and improve treatment outcomes. Antibiotic treatment significantly reduced lung injury, lowered inflammatory markers such as interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α), and improved lung structure in the AP model [26]. Lastly, antibiotics have been proven to minimize com- plications associated with AP, such as sepsis and acute respiratory distress syndrome.

(ARDS), by inhibiting the production of bacterial toxins and modulating the inflam- matory response [27]. Therefore, antibiotics, as an important treatment option for improving the prognosis of critically ill patients in the ICU, can reduce the in-hospital mortality rate of patients with AP when used rationally.

Based on the significant differences between the antibiotic use and non-use groups, factors such as gender, age, Body Mass Index (BMI), blood sugar, number of comor- bidities, and time of ICU admission showed no statistical significance between the two groups (p > 0.05). Other parameters like PO2, PCO2, pH, and blood potassium are general blood gas characteristics of patients. There may be significant differences in these features before and after the progression to AP. This suggests that in studying whether these indicators are directly related to in-hospital mortality in patients with AP, further analysis of the changes in these indicators is necessary, as conclusions drawn from a single test may not be reliable. Indicators with statistically significant differences (p < 0.05), such as hemoglobin, white blood cells, lymphocytes, monocytes, neutrophils; as well as complications like congestive heart failure, cerebrovascular disease, myocardial infarction, and malignant tumors; along with the SOFA score reflecting the type and severity of the illness at admission, suggest that diseases in the antibiotic use group were more complex and severe compared to the non-antibiotic group, consistent with clinical experience. BMI is a unique indicator. According to our study results (see Additional file 6: Figure S2), in the group with BMI > = 18.5, the use of antibiotics significantly reduced the in-hospital mortality rate of patients with AP (OR 0.40, 95% CI 0.24–0.67, P < 0.001); however, this was not statistically sig- nificant in the group with BMI < 18.5. This is consistent with the’obesity paradox’ [28], suggesting that patients with a higher BMI at admission may better support the body’s nutritional needs during illness, thus having a higher probability of survival.

The analysis results indicated that the most commonly used medications for patients with AP arevancomycin and cephalosporins, which are consistent with related research findings in the literature [29,30,31]. This indicates that current clinical drug interventions for AP patients in the United States (including diagnostic and treatment protocols) are relatively consistent with practices in other countries. The results of this study show that the main pathogen composition inpatients with AP is consistent with findings from other studies [5, 30, 32]. This suggests a possible correlation between mul- tiple pathogens that invade the respiratory tract (such as Candida, coagulase-positive Staphylococcus aureus, and Pseudomonas aeruginosa) and gastrointestinal colonizing microorganisms [5, 6, 33]. These pathogens are also the main cause of lung injury. In particular, the study [34] have shown that uncommon microorganisms, such as Acine- tobacter and Aeromonas, are also pathogenic causes of AP. In the 1970s, anaerobic bacteria were considered the main pathogens in AP. However, recent studies indicate that the detection rate of anaerobes in AP patients has significantly declined, while the detection rate of gram-negative bacilli has increased [35]. Other study [36]have pointed out that the isolation rate of gram-negative bacilli has significantly increased among severe AP patients, while the isolation rate of anaerobes has decreased. Furthermore, the study [5] indicates that in hospitalized severe AP patients, gram-negative bacte- ria (such as Escherichia coli, Klebsiella pneumoniae, Serratia, and Proteus species),

Staphylococcus aureus, and Streptococcus pneumoniae are the most common microor- ganisms. Notably, ICU patients who have had strokes are more likely to experience dysphagia and reflux due to the severity of their condition compared to patients in traditional wards. This may lead to an increased incidence of aspiration pneumonia, thus exacerbating lung damage. Therefore, clinical healthcare providers should pay special attention to this high-risk population.

Vancomycin is a glycopeptide antibiotic widely used to treat infections caused by pathogens such as Staphylococcus aureus and Streptococcus pneumoniae [37,38,39]. It is more effective than cephalosporin antibiotics, has no cross-resistance with other antibiotics [40]. This study also found that no strains showed resistance to vancomycin. Furthermore, according to the results of this paper, most patients are simultaneously infected with both Gram-positive and Gram-negative bacteria, thus the combined use of vancomycin and piperacillin-tazobactam aligns with the prescribing habits of clinical doctors.

Each generation of cephalosporins targets different pathogens [41,42,43,44]. FirGC was developed between 1962 and 1970. FirGC’s antibacterial spectrum is effective against Gram-positive bacteria but less so against streptococci, with weak blood–brain barrier penetration, making it unsuitable for central infections. Second-generation cephalosporins, mostly developed between 1970 and 1976, have better antibacterial activity against Gram-negative bacteria than FirGC. The antibacterial spectrum of third-generation cephalosporins is expanded from the second generation but has a weaker effect on Gram-positive bacteria compared to FirGC. Recently developed ForGC has a broader antibacterial spectrum, not only effective against Gram-negative bacteria but also resistant to Staphylococcus aureus.

Considering the results of this study and the characteristics of different antimicro- bial drugs, patients are most commonly treated with a combination of vancomycin and piperacillin-tazobactam. This may be because these two drugs have a broad antibac- terial spectrum and are acceptable to more patients [37, 45]. Therefore, it is the first choice for clinical antibiotic treatment in patients with AP infected with different bac- teria or more complex microbial communities. Patients treated with levofloxacin + metronidazole have the highest survival rate due to the broad antibacterial spectrum of levofloxacin, especially its effectiveness against Gram-negative bacteria, but it has weaker effects on some Gram-positive and anaerobic bacteria [46]. Metronidazole is particularly effective against infections caused by anaerobic bacteria (such as Clostrid- ium species) and certain protozoa (such as Trichomonas) [47].At the same time, these patients have milder conditions and clearer bacterial infections, hence the combined treatment has the highest cure rate. However, the survival rate of patients treated with ForGC + Van is very low. This reminds us that to maximize the survival rate of patients with AP, we need to be aware of the occurrence of drug misuse and administer the drugs accurately at the appropriate time.

Therefore, we suggest that the early treatment of patients with AP should involve the use of more adaptable broad-spectrum antibiotics based on the patient’s clinical symptoms. In the later stages, as understanding of the pathogen’s resistance increases, more sensitive antibiotics should be used. In clinical practice, doctors should develop a more personalized and precise drug intervention program for patients and adjust the dose and duration of antibiotics in a timely manner, in addition to the rational use of MV according to the patient’s clinical indicators in order to improve the recovery of patients with aspiration pneumonia.

Inflammation-related indicators of patients can assist in diagnosing the occurrence of AP and the rational use of medications [35, 48, 49]. However, the results of this study did not show a significant increase in the median and IQR values of inflammatory markers in the medication group compared to the untreated group. This could be attributed to the high sensitivity but low specificity of the inflammation response index. No significant increase was observed in the median and IQR values. This may be because the data used came from the first ICU admission of patients with AP, where the inflammatory response was still in its initial stage. Although there was a statistically significant difference between the groups, there was no significant change in the values of the inflammatory markers.

The results of this study on pathogen resistance indicate that the pathogens in the included AP patients have poor resistance. This means that antibiotic treatment is appropriate and effective, impacting the mortality rate of AP patients. This further suggests that the medication experience of the medical institutions or centers involved in the study can serve as a reference for other medical institutions.

This study inevitably has some limitations. First, as the MIMIC-IV database largely comprises Caucasian patients, the findings may require validation across diverse populations to account for genetic variations. Additionally, we did not exam- ine potential causative factors for aspiration pneumonia (AP) or prognostic variables influencing AP outcomes. Furthermore, due to the study’s retrospective design,numer- ous confounding factors must be controlled through methods such as Propensity Score Matching (PSM), Inverse Probability Weighting (IPW), or multivariate adjustment. Despite using PSM to adjust for confounding variables, the possibility of residual unmeasured confounders and selection bias remains. Moreover, AP case identification relied solely on ICD codes. Although ICD codes facilitate large-scale data extrac- tion, they may introduce diagnostic bias, as AP is often clinically diagnosed based on symptoms, imaging findings, and patient history rather than standardized diagnostic criteria. Exclusive reliance on ICD codes could overestimate AP case numbers, par- ticularly when direct clinical evidence (e.g., respiratory failure, new-onset opacities post-vomiting) is lacking. Readers are encouraged to consider this potential overdiag- nosis when interpreting the study’s findings. Finally, patients who used antibiotics, although they had similar baseline characteristics after PSM, the quality of outpatient care or other macro healthcare may differ. These issues will be one of the main focuses of our next study.

Conclusion

Antibiotics use is closely associated with lower in-hospital mortality in ICU patients with AP. Furthermore,this study describes the basic clinical characteristics, pathogen composition, resistance, and antibiotic use in patients with AP in the ICU from the MIMIC-IV database. The study results provide a reference path and theoretical basis for formulating more rational medication intervention measures for patients with AP.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Abbreviations

AP:

Aspiration pneumonia

ICU:

Intensive care units

MIMIC-IV:

Medical Information Mart for Intensive Care IV

PSM:

Propensity score matching

IPW:

Propensity score- based inverse probability weighting

OR:

Odds ratio

SQL:

Structured query language

CI:

Confdence interval

IQR:

Interquartile range

SOFA:

Sequential Organ Failure Assessment

SMD:

Standardized mean diference

GBM:

Gradient boosting model

INR:

International normalized ratio

pH:

Potential of Hydrogen

APSIII:

Acute physi- ology score –III

FirGC:

First-Generation Cephalosporins

ThirGC:

Third-Generation Cephalosporins

ForGC:

Fourth-Generation Cephalosporins

BMI:

Body Mass Index

IL-6:

Interleukin-6

TNF-α:

Tumor necrosis factor-α

ARDS:

Acute respiratory distress syndrome

References

  1. Lindenauer PK, Strait KM, Grady JN, Ngo CK, Parisi ML, Metersky M, et al. Variation in the diagnosis of aspiration pneumonia and association with hospital pneumonia outcomes. Ann Am Thorac Soc. 2018;15(5):562–9.

    Article  PubMed  Google Scholar 

  2. Yoon HY, Shim SS, Kim SJ, Lee JH, Chang JH, Lee SH, et al. Long-term mor- tality and prognostic factors in aspiration pneumonia. J Am Med Dir Assoc. 2019;20(9):1098–104.

    Article  PubMed  Google Scholar 

  3. Rodriguez AE, Restrepo MI. New perspectives in aspiration community acquired Pneumonia. Expert Rev Clin Pharmacol. 2019;12(10):991–1002.

  4. Metlay JP, Waterer GW, Long AC, Anzueto A, Brozek J, Crothers K, et al. Diagnosis and treatment of adults with community-acquired pneumonia. An offi- cial clinical practice guideline of the American Thoracic Society and Infectious Diseases Society of America. Amer J Respir Crit Care Med. 2019;200(7):e45–67.

    Article  Google Scholar 

  5. Niederman MS, Cilloniz C. Aspiration pneumonia. Revista EspaËœnola de Quimioterapia. 2022;35(Suppl 1):73.

  6. Berger SC. An introduction to aspiration pneumonia. US Pharmacist. 2021;46(7).

  7. Bowerman TJ, Zhang J, Waite LM. Antibacterial treatment of aspiration pneu- monia in older people: a systematic review. Clinical interventions in aging. 2018;p. 2201–2213.

  8. Song L, Li M, Zhang T, Huang L, Ying J, Ying L. Association between high- flow nasal cannula use and mortality in patients with sepsis-induced acute lung injury: a retrospective propensity score-matched cohort study. BMC Pulm Med. 2024;24(1):197.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Johnson A, Bulgarelli L, Pollard T, Horng S, Celi LA, Mark R. MIMIC-IV. Phy- sioNet Available online at: https://physionet.org/content/mimiciv/22/ (accessed Jan 6, 2023).

  10. Johnson AE, Bulgarelli L, Shen L, Gayles A, Shammout A, Horng S, et al. MIMIC-IV, a freely accessible electronic health record dataset. Scientific data. 2023;10(1):1.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Lu X, Wang X, Gao Y, Yu S, Zhao L, Zhang Z, et al. Efficacy and safety of corticosteroids for septic shock in immunocompromised patients: A cohort study from MIMIC. Am J Emerg Med. 2021;42:121–6.

    Article  PubMed  Google Scholar 

  12. Feng M, McSparron JI, Kien DT, Stone DJ, Roberts DH, Schwartzstein RM, et al. Transthoracic echocardiography and mortality in sepsis: analysis of the MIMIC-III database. Intensive Care Med. 2018;44:884–92.

    Article  CAS  PubMed  Google Scholar 

  13. Van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained equations in R. J Stat Softw. 2011;45:1–67.

    Article  Google Scholar 

  14. WANG Wei YF. A generalized linear model based on multiple imputation in kidney disease research. Chinese Journal of Nephrology. Dialysis & Transplantation. 2021;30(5):476–479.

  15. Schober P, Vetter TR. Propensity score matching in observational research. Anesth Analg. 2020;130(6):1616–7.

    Article  PubMed  Google Scholar 

  16. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivar Behav Res. 2011;46(3):399–424.

    Article  Google Scholar 

  17. Graffeo N, Latouche A, Le Tourneau C, Chevret S. ipcwswitch: an R package for inverse probability of censoring weighting with an application to switches in clinical trials. Comput Biol Med. 2019;111: 103339.

    Article  PubMed  Google Scholar 

  18. Khan S, Ugander J. Adaptive normalization for IPW estimation. Journal of Causal Inference. 2023;11(1):20220019.

    Article  Google Scholar 

  19. Zhu Y, Coffman DL, Ghosh D. A boosting algorithm for estimating general- ized propensity scores with continuous treatments. Journal of causal inference. 2015;3(1):25–40.

    Article  PubMed  Google Scholar 

  20. GuolinKe QM, Finley T, Wang T, Chen W, Ma W, Ye Q, et al. Lightgbm: A highly efficient gradient boosting decision tree. Adv Neural Inf Process Syst. 2017;30:52.

  21. Louk MHL, Tama BA. Dual-IDS: A bagging-based gradient boosting decision tree model for network anomaly intrusion detection system. Expert Syst Appl. 2023;213: 119030.

    Article  Google Scholar 

  22. Sa MP, Jacquemyn X, Van den Eynde J, Tasoudis P, Dokollari A, Torregrossa G, et al. Impact of prosthesis-patient mismatch after transcatheter aortic valve replacement: meta-analysis of Kaplan-Meier–derived individual patient data. Cardiovascular Imaging. 2023;16(3):298–310.

    PubMed  Google Scholar 

  23. Nakagawa Y, Kato H, Iwamoto T, et al. Investigation of antimicrobial therapy for aspiration pneumonia in elderly nursing homes: a retrospective observational study in Japan. Pharmacy Practice. 2023;21(3):1–10.

    Article  Google Scholar 

  24. Binz J, Heft M, Robinson S, Jensen H, Newton J. Utilizing procalci- tonin in a clinical setting to help differentiate between aspiration pneumonia and aspiration pneumonitis. Diagn Microbiol Infect Dis. 2023;105(1): 115821.

    Article  PubMed  Google Scholar 

  25. Smith S, Rowbotham NJ. Inhaled anti-pseudomonal antibiotics for long-term therapy in cystic fibrosis. Cochrane Database Syst Rev. 2022;11(11):CD001021.

  26. Hashimoto Y, Eguchi A, Wei Y, Shinno-Hashimoto H, Fujita Y, Ishima T, et al. Antibiotic-induced microbiome depletion improves LPS-induced acute lung injury via gut-lung axis. Life Sci. 2022;307: 120885.

    Article  CAS  PubMed  Google Scholar 

  27. Sauer A, Peukert K, Putensen C, Bode C. Antibiotics as immunomodu- lators: a potential pharmacologic approach for ARDS treatment. Eur Respir Rev. 2021;30(162):210093.

  28. Lv Y, Mao C, Gao X, Ji JS, Kraus VB, Yin Z, et al. The obesity paradox is mostly driven by decreased noncardiovascular disease mortality in the oldest old in China: a 20-year prospective cohort study. Nature Aging. 2022;2(5):389–96.

    Article  PubMed  Google Scholar 

  29. Tabatabaei M, Dorvashy G, Soltani R, Samsamshariat S, Meamar R, Sabzghabaee AM. The safety and effectiveness of pharmacotherapy for opioid-overdose induced aspiration pneumonia in a referral poisoning management University hospital in Iran. J Res Pharm Pract. 2021;10(1):38.

    Article  PubMed  PubMed Central  Google Scholar 

  30. Finegold SM. Aspiration pneumonia. In: Seminars in respiratory and critical care medicine. vol. 16. Copyright© 1995 by Thieme Medical Publishers, Inc.; 1995. p. 475–483.

  31. Moore FA. Treatment of aspiration in intensive care unit patients. J Parenter Enter Nutr. 2002;26:S69–74.

    Article  Google Scholar 

  32. Rebuck JA, Rasmussen JR, Olsen KM. Clinical aspiration-related practice pat- terns in the intensive care unit: a physician survey. Crit Care Med. 2001;29(12):2239–44.

    Article  CAS  PubMed  Google Scholar 

  33. Finegold SM. Aspiration pneumonia. Reviews of infectious diseases. 1991;13(Sup- plement 9):S737–S742.

  34. Patil V, Noronha V, Shrirangwar S, Menon N, Abraham G, Chandrasekharan A, et al. Aspiration pneumonia in head and neck cancer patients undergoing concurrent chemoradiation from India: findings from a post hoc analysis of a phase 3 study. Cancer Med. 2021;10(19):6725–35.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Marik PE. Aspiration pneumonitis and aspiration pneumonia. N Engl J Med. 2001;344(9):665–71.

    Article  CAS  PubMed  Google Scholar 

  36. El-Solh AA, Pietrantoni C, Bhat A, Aquilina AT, Okada M, Grover V, et al. Microbiology of severe aspiration pneumonia in institutionalized elderly. Ameri- can journal of respiratory and critical care medicine. 2003;167(12):1650–4.

    Article  Google Scholar 

  37. Deora H, Nagesh M, Garg K, Singh M, Chandra SP, Kale SS. Topical Van- comycin for Prevention of Surgical Site Infection in Cranial Surgeries: Results of an Updated Systematic Review. Meta-Analysis and Meta-Regression Neurology India. 2023;71(5):875–83.

    PubMed  Google Scholar 

  38. Bosheva M, Gujabidze R, Karoly EE, Nemeth A, Saulay M, Smart JI, et al. Aphase 3, randomized, investigator-blinded trial comparing ceftobiprole with a standard-of-care cephalosporin, with or without vancomycin, for the treatment of pneumonia in pediatric patients. Pediatr Infect Dis J. 2021;40(6):e222.

    Article  PubMed  PubMed Central  Google Scholar 

  39. Hoover A, Zimmerman JA, Wiese S, Modi A. Evaluation of Empiric Vancomycin for Fevers During High-dose Cytarabine Administration. J Pediatr Hematol Oncol. 2021;43(7): e1010.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. van der Veen A, Somers A, Vanhaesebrouck S, Ter Heine R, Bruggemann R, Allegaert K, et al. Variation in vancomycin dosing and therapeutic drug moni- toring practices in neonatal intensive care units. Int J Clin Pharm. 2022;44(2):564–9.

    Article  PubMed  Google Scholar 

  41. Mor´an-D´ıaz JR, Neveros-Ju´arez ·F, Arellano-Mendoza MG, Quintana-Zavala D, Lara-Salazar O, Trujillo-Ferrara JG, et al. QSAR analysis of five generations of cephalosporins to establish the structural basis of activity against methicillin- resistant and methicillin-sensitive Staphylococcus aureus. Molecular Diversity. 2023;p. 1–17.

  42. Abdelraouf K, Chavda KD, Satlin MJ, Jenkins SG, Kreiswirth BN, Nico-lau DP. Piperacillin-tazobactam-resistant/third-generation cephalosporin- susceptible Escherichia coli and Klebsiella pneumoniae isolates: resistance mech- anisms and in vitro-in vivo discordance. Int J Antimicrobial Agents. 2020;55(3):105885.

    Article  CAS  Google Scholar 

  43. Versporten A, Bruyndonckx R, Adriaenssens N, Hens N, Monnet DL, Molen-Berghs G, et al. Consumption of cephalosporins in the community, Euro-pean Union/European Economic Area. J Antimicrob Chemother. 2021;76(Supplement 2):ii22–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. Fong I. New Cephalosporins: Fifth and Sixth Generations. In: New Antimicro- bials: For the Present and the Future. Springer; 2023. p. 25–38.

  45. Gin A, Dilay L, Karlowsky JA, Walkty A, Rubinstein E, Zhanel GG. Piperacillin– tazobactam: a β-lactam/β-lactamase inhibitor combination. Expert Rev Anti Infect Ther. 2007;5(3):365–83.

    Article  CAS  PubMed  Google Scholar 

  46. Croom KF, Goa KL. Levofloxacin: a review of its use in the treatment of bacterial infections in the United States. Drugs. 2003;63:2769–802.

    Article  CAS  PubMed  Google Scholar 

  47. Gharebaghi R, Heidary F, Moradi M, Parvizi M. Metronidazole; a Potential Novel Addition to the COVID-19 Treatment Regimen. Arch Acad Emerg Med. 2020;8(1):e40.

  48. Komiya K, Rubin BK, Kadota JI, Mukae H, Akaba T, Moro H, et al. Prognostic implications of aspiration pneumonia in patients with community acquired pneumonia: A systematic review with meta-analysis. Sci Rep. 2016;6(1):38097.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Cao W, Ni X, Gan M, Xie B, Xie Y, Wang Q, et al. Treatment and diagnosis of hyperlipidemia acute pancreatitis in pregnancy associated with pre-pregnancy obesity and diabetes: A case report. Exp Ther Med. 2023;26(6):1–8.

    Article  Google Scholar 

Download references

Acknowledgements

We appreciate the researchers at the MIT Laboratory for Computational Physiology for publicly sharing of the MIMIC-IV clinical database.

Funding

This work was supported by the National Key Research and Development Pro- gram of China (No.2020AAA0109703), the Key Scientific Research Project of Higher Education Institutions in Henan Province (No.24A520058,No.24A520060, No.23A520022), the Postgraduate Education Reform and Ouality Improvement Project of Henan Province(No.YJS2024AL053).

Author information

Author notes

  1. Di Zhang and Guan Yang contributed equally to this work.

Authors and Affiliations

  1. School of Computer Science, Zhongyuan University of Technology, Zhengzhou, 450007, China

    Di Zhang, Guan Yang, Xingang Hu, Xiaoming Liu & Aojun Zhang

  2. Department of Respiratory and Critical Care Medicine, Henan Provincial People’s Hospital, Zhengzhou, 450000, China

    Xingang Hu & Jie Zhang

  3. Zhengzhou Key Laboratory of Text Processing and Image Understanding, Zhengzhou, State, 450007, China

    Guan Yang

  4. China Language Intelligence Research Center, Beijing, 100000, China

    Xiaoming Liu

  5. Medical Department, Kaifeng University, Kaifeng, 475004, China

    Dongqing Jia

Authors
  1. Di Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Guan Yang
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Xingang Hu
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Xiaoming Liu
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Jie Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Dongqing Jia
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Aojun Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

DZ: conducted a literature search, conceptualized data analysis, illustrations, manuscript and citations. XH: proposed the study, conducted a literature search, notified additional corrections and clarifications outlined the inclusion and exclusion criteria for downloading and organizing the papers, reviewed and edited the study. GY: proposed the study, conducted a literature search and completed a fulltext review, finding gaps, challenges and future directions, and then reviewed and edited the study. XL, JZ, DJ and AZ: Reviewed and edited the study. All authors have agreed to be personally accountable for their contributions. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Guan Yang or Xingang Hu.

Ethics declarations

Ethics approval and consent to participate

The MIMIC-IV databases have received ethical approval from the Institutional Review Boards at Beth Israel Deaconess Medical Center and Massachusetts Insti- tute of Technology. As the databases do not contain protected health information, a waiver of informed consent was included in the approval from the Institutional Review Boards at Beth Israel Deaconess Medical Center and Massachusetts Insti- tute of Technology. Therefore, this manuscript does not involve a research protocol requiring approval by the relevant institutional review board or ethics committee. All methods in this study were carried out in accordance with relevant guidelines and regulations (declarations of Helsinki).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Table S1: Percentage of missing data in the variables in the interest.

12890_2024_3441_MOESM2_ESM.pdf

Additional file 2: Figure S1: Relative infuence factor of covariates. The relative infuence factor measures how discriminative the 35 covariates of the propensity score model are when predicting the likelihood of antibiotic usage.

12890_2024_3441_MOESM3_ESM.xls

Additional file 3: Table S2: Resistance of main gram-negative bacteria to different antibacterial drugs in respiratory tract specimens of AP patients treated with vancomycin combined with Piperacillin-Tazobactam.

12890_2024_3441_MOESM4_ESM.xls

Additional file 4: Table S3: Distribution and Proportional Composition of Main Pathogenic Bacteria in Respiratory Tract Specimens from Patients with Aspira-tion Pneumonia Treated with levofloxacin combined with metronidazole.

12890_2024_3441_MOESM5_ESM.xls

Additional file 5: Table S4: Resistance of main gram-positive bacteria to differ-ent antibacterial drugs in respiratory tract specimens of Aspiration Pneumonia patients treated with levofloxacin combined with metronidazole.

Additional file 6: Figure S2: Association between antibiotic use and in-hospital morality in subgroup.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhang, D., Yang, G., Hu, X. et al. Antibiotics versus Non-Antibiotic in the treatment of Aspiration Pneumonia: analysis of the MIMIC-IV database. BMC Pulm Med 24, 621 (2024). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12890-024-03441-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12890-024-03441-8

Keywords

BMC Pulmonary Medicine

ISSN: 1471-2466

Contact us