The incidence and characteristics of ventilator-associated pneumonia in a regional nontertiary Australian intensive care unit: A retrospective clinical audit study

Machine learning prediction models for multidrug-resistant organism infections in ICU ventilator-associated pneumonia patients: Analysis using the MIMIC-IV database

OBJECTIVE

This study aims to construct and compare four machine learning models using the MIMIC-IV database to identify high-risk factors for multidrug-resistant organism (MDRO) infection in Ventilator-associated pneumonia (VAP) patients.

METHODS

The study included 972 VAP patients from the MIMIC-IV database. Data encompassing demographic information, vital signs, laboratory results, and other relevant variables were collected. The class imbalance issue was addressed using the Synthetic Minority Over-sampling Technique (SMOTE). The dataset was randomly split into training and testing sets (8:2). LASSO regression and feature importance scores were used for feature selection. Clinical prediction models were built using logistic regression, XGBoost, random forest and gradient boosting machine. The performance of the models was evaluated through receiver operating characteristic(ROC) curve analysis.Model calibration was assessed using calibration curves and Brier scores. The effectiveness was evaluated through Decision Curve Analysis (DCA). SHAP was utilized for model interpretation.

RESULTS

Among 972 patients, 824 were non-MDROs-VAP and 128 were MDROs-VAP. Comparative analysis revealed statistically significant differences in various clinical parameters. XGBoost exhibited the best predictive performance, incorporating 20 features with an AUC of 0.831 (95 % CI: 0.785-0.877) on the test set. Calibration curves demonstrated robust consistency, corroborated by Decision Curve Analysis (DCA) affirming the clinical utility. SHAP analysis identified the most important features: red cell distribution width, duration of mechanical ventilation, anion gap, basophil percentage, and neutrophil percentage.

CONCLUSION

This study established and compared four machine learning models for MDROs infections in VAP patients. XGBoost was identified as the optimal predictor, and SHAP values provided insights into 20 independent risk factors, confirming its excellent predictive value.

IMPLICATIONS FOR CLINICAL PRACTICE

VAP is a common infection in ICU patients with a heightened risk of MDRO and increased mortality. The recognition of high bias in existing models calls for future research to employ rigorous methodologies and robust data sources, aiming to develop and validate more accurate and clinically applicable predictive models for MDROs infections in VAP patients.

Bacteriology of endotracheal tube biofilms and antibiotic resistance: a systematic review

Bacteria commonly adhere to surfaces and produce polymeric material to encase the attached cells to form communities called biofilms. Within these biofilms, bacteria can appear to be many times more resistant to antibiotics or disinfectants. This systematic review explores the prevalence and microbial profile associated with biofilm production of bacteria isolated from endotracheal tubes and its associations with antimicrobial resistance. A comprehensive search was performed on databases PubMed, Embase, and Google Scholar for relevant articles published between 1st January 2000 and 31st December 2022. The relevant articles were exported to Mendeley Desktop 1.19.8 and screened by title and abstract, followed by full text screening based on the eligibility criteria of the study. Quality assessment of the studies was performed using the Newcastle-Ottawa Scale (NOS) customized for cross-sectional studies. Furthermore, the prevalence of antimicrobial resistance in biofilm-producers isolated from endotracheal tube specimens was investigated. Twenty studies encompassing 981 endotracheal tubes met the eligibility criteria. Pseudomonas spp. and Acinetobacter spp. were predominant isolates among the biofilm producers. These biofilms provided strong resistance against commonly used antibiotics. The highest resistance rate observed in Pseudomonas spp. was against fluoroquinolones whereas the least resistance was seen against piperacillin-tazobactam. A similar trend of susceptibility was observed in Acinetobacter spp. with a very high resistance rate against fluoroquinolones, third-generation cephalosporins and carbapenems. In conclusion, endotracheal tubes were associated with colonization by biofilm forming bacteria with varying levels of antimicrobial resistance. Biofilms may promote the occurrence of recalcitrant infections in endotracheal tubes which need to be managed with appropriate protocols and antimicrobial stewardship. Research focus should shift towards meticulous exploration of biofilm-associated infections to improve detection and management.

Identifying the barriers and facilitators to implementation of ventilator bundle in the nepalese intensive care unit: A descriptive qualitative study

BACKGROUND

A ventilator bundle is an effective preventive strategy against the development of ventilator-associated pneumonia (VAP). However, in clinical practice ventilator bundle implementation is poor. Understanding the barriers to ventilator bundle implementation in low- and middle-income countries can inform the development of effective implementation strategies to reduce the burden of VAP.

OBJECTIVES

The primary objective of this study was to explore the barriers and facilitators of ventilator bundle implementation perceived by healthcare professionals (HCPs) working in intensive care units (ICU) in Nepal. The secondary objective was to prioritise the barriers when developing implementation strategies.

METHODS

This study used a pragmatic approach comprising a series of methods to identify the implementation strategies: (i) Barriers and facilitators were explored using a qualitative study design. Twenty-one HCPs selected using the maximum variation sampling technique from a large tertiary hospital, completed semistructured interviews. All the interviews were recorded, transcribed word-by-word, and uploaded into NVivo for analysis using the thematic analysis approach. (ii) After analysis, nine participants were selecteded to determine the priority order of the barriers using a barrier identification and mitigation tool.

RESULTS

The data analysis revealed five main themes and 19 subthemes that affected ventilator bundle implementation. The main themes were provider-related factors, organisational and practice-related factors, performances of work, environmental conditions, and patient-related factors. The common barriers were job insecurity, poor knowledge, negative attitude, insufficient equipment, and severity of patient disease. Common facilitators were educational training, equipment functioning, adequate staff, strong leadership, and organisational support. Finally, eight main barriers were prioritised to target the change.

CONCLUSION

The barriers to implementing ventilator bundles in ICUs were identified. Focussing on addressing the prioritised barriers may aid in improving patient care and safety in ICUs. Results may guide HCPs in the development of implementation strategies to reduce the burden of VAP.

Machine learning algorithms for prediction of ventilator associated pneumonia in traumatic brain injury patients from the MIMIC-III database

BACKGROUND

Ventilator associated pneumonia (VAP) is a common complication and associated with poor prognosis of traumatic brain injury (TBI) patients.

OBJECTIVES

This study was conducted to explore the predictive performance of different machine-learning algorithms for VAP in TBI patients.

METHODS

TBI patients receiving mechanical ventilation more than 48 hours from the Medical Information Mart for Intensive Care-III (MIMIC-III) database were eligible for the study. The VAP was confirmed based on the ICD-9 code. Included patients were separated to the training cohort and the validation cohort with a ratio of 7:3. Predictive models based on different machine learning algorithms were developed using 5-fold cross validation in the training cohort and then verified in the validation cohort by evaluating the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, accuracy and F score.

RESULTS

786 TBI patients from the MIMIC-III were finally included with the VAP incidence of 44.0%. The random forest performed the best on predicting VAP in the training cohort with a AUC of 1.000. The XGBoost and AdaBoost were ranked the second and the third with a AUC of 0.915 and 0.789 in the training cohort. While the AdaBoost performed the best on predicting VAP in the validation cohort with a AUC of 0.706. The XGBoost and random forest were ranked the second and the third with the AUC of 0.685 and 0.683 in the validation cohort. Generally, the random forest and XGBoost were likely to be over-fitting while the AdaBoost was relatively stable in predicting the VAP.

CONCLUSIONS

The AdaBoost performed well and stably on predicting the VAP in TBI patients. Developing programs using AdaBoost in portable electronic devices may effectively assist physicians in assessing the risk of VAP in TBI.

The incidence, characteristics and in-hospital mortality of non-ventilator-associated hospital-acquired pneumonia in regional Queensland: A retrospective descriptive study

OBJECTIVE

The aim of this study was to determine the incidence, characteristics and in-hospital mortality of non-ventilator-associated hospital-acquired pneumonia (NV-HAP) in a regional (Modified Monash Model 2) Australian hospital.

METHODS

All cases with NV-HAP were obtained from the Business Analysis and Decision Support (BADS) Unit between 1st January 2013 and 31st December 2018. Medical records were reviewed, and data pertaining to incidence, characteristics (age and gender), length of stay, co-morbidities (measured using the Charlson Comorbidity Index) and in-hospital mortality were extracted. Incidence rate was calculated as a proportion of NV-HAP cases per 1000 bed-days.

DESIGN

A retrospective study design was used to review all cases of NV-HAP between 1 January 2013 and 31 December 2018 at a single regional Australian hospital. Using the Modified Monash Model (MMM), our regional setting is classified as a regional centre (MMM-2).

SETTING

Rockhampton Hospital, Australia.

PARTICIPANTS

Patient cases.

MAIN OUTCOME MEASURES

Incidence rate, Incidence proportion, mortality.

RESULTS

A total of 501 cases were identified with an incidence rate of 0.98 cases per 1000 bed-days over the study period 2013-2018. Cases with NV-HAP had a median age of 78.2 years (interquartile range 18.8), a median length of stay of 13.0 days (interquartile range 12.0) and a median Charlson Comorbidity Index score of 3.0 out of 39 (interquartile range 3.0), and a greater proportion was male (n = 297, 57%). The in-hospital mortality rate for NV-HAP cases was 18.9%.

CONCLUSION

This study revealed an overall incidence rate of 0.98 cases per 1000 bed-days from 2013 to 2018 in a regional Australian hospital. In addition, this study provided the descriptive characteristics for patients with NV-HAP at our regional hospital.

Efficacy of comprehensive unit-based safety program to prevent ventilator associated-pneumonia for mechanically ventilated patients in China: A propensity-matched analysis

Background

Ventilator-associated pneumonia (VAP) is the most common healthcare-associated infection (HAI) in patients with mechanical ventilation. VAP is largely preventable, and a comprehensive unit-based safety program (CUSP) has effectively reduced HAI. In this study, we aim to comprehensively investigate the effect of implementing the CUSP in patients requiring mechanical ventilation.

Methods

In this uncontrolled before-and-after trial conducted in two intensive care unit (ICU) settings in China, patients requiring invasive mechanical ventilation were enrolled. Patients were divided into two groups based on the implementation of CUSP. The primary outcome was the incidence of VAP. The secondary outcomes were the time from intubation to VAP, days of antibiotic use for VAP treatments, rate of other infection, length of stay (LOS) in ICU, hospital LOS, and safety culture score. Joinpoint regression analysis was used to test the changes in trends of VAP rate for statistical significance. Propensity score matching (1:1 matching) was used to reduce the potential bias between CUSP and no CUSP groups. Univariate and multivariate logistic/linear regression analyses were performed to evaluate the association between the use of CUSP and clinical outcomes. This study was registered at the Chinese Clinical Trial Registry (chictr.org.cn), registration number: ChiCTR1900025391.

Results

A total of 1,004 patients from the transplant ICU (TICU) and 1,001 patients from the surgical ICU (SICU) were enrolled in the study from January 2016 to March 2022. Before propensity score matching, the incidences of VAP decreased from 35.1/1,000 ventilator days in the no CUSP group to 12.3/1,000 ventilator days in the CUSP group in the TICU setting (adjusted odds ratio [OR], 0.30; 95% confidence interval [CI], 0.15-0.59). The results of the joinpoint regression analysis confirmed that the implementation of CUSP significantly decreased the incidences of VAP. After propensity score matching in TICU setting, the CUSP group reported a lower incidence of VAP (30.4 vs. 9.7‰, P = 0.003; adjusted OR = 0.26, 95% CI: 0.10-0.76), lower wound infection (3.4 vs. 0.9%, P = 0.048; adjusted OR = 0.73, 95% CI: 0.50-0.95), shorter ICU LOS [3.5(2.3-5.3) vs. 2.5(2.0-4.5) days; P = 0.003, adjusted estimate = -0.34, 95% CI: -0.92 to -0.14], and higher safety culture score (149.40 ± 11.74 vs. 153.37 ± 9.74; P = 0.002). Similar results were also observed in the SICU setting between the no CUSP and CUSP group.

Conclusions

The implementation of CSUP for patients receiving mechanical ventilation could significantly reduce the incidences of VAP, and other infections, prolong the time until the VAP occurrence, reduces the days of antibiotic use for VAP, shorten the ICU and hospital LOS, and enhance the awareness of safety culture.