Use of computerized surveillance to detect nosocomial pneumonia in neonatal intensive care unit patients

The informatics pathway for hospital infection quality control monitoring

Objective

To analyze the results and current status of hospital infection monitoring and quality control work at a tertiary specialized cancer hospital, providing references for further improvement of HAI monitoring quality.

Methods

Combined information-based monitoring with manual quality control measures for hospital infection surveillance, collecting data such as hospital infection report cards and missed reports to analyze the current status of hospital infection monitoring. Starting from January 2023, further comprehensive quality control measures were taken, and data before and after the implementation were compared to evaluate the effectiveness of the quality control work.

Results

After the implementation of comprehensive management measures, the incidence rate of hospital infection in the hospital decreased from 0.60 to 0.52%, the correct rate of early warning disposal increased from 85.31 to 89.53%, and the handling rate within 24 h of early warning increased from 20.51 to 72.99%. All these differences are statistically significant (p < 0.05). After the comprehensive management measures were adopted, more missed cases were detected and reported, resulting in an increase in the missed reporting rate from 5.36 to 12.33% (p < 0.05). While the number of reports increased, the accuracy of the reports decreased from 67.54 to 54.05% (p < 0.05).

Conclusion

Information systems can enhance the efficiency of hospital infection surveillance through real-time monitoring and automatic early warning, improve the quality of reporting, and thereby contribute to the reduction of the incidence rate of hospital infections. However, the quality of monitoring is still influenced by human factors such as whether the early warning rules are scientifically set and whether the determination of hospital infections is accurate. There may be situations where the missed reporting rate is underestimated and the quality of the report cards is not high. This indicates that while adopting information-based monitoring, we cannot ignore the management of quality control. It is necessary to continuously strengthen the investigation of missed reporting, improve the quality of report cards, and ensure the authenticity and accuracy of the results of hospital infection monitoring.

Automated surveillance of non-ventilator-associated hospital-acquired pneumonia (nvHAP): a systematic literature review

BACKGROUND

Hospital-acquired pneumonia (HAP) and its specific subset, non-ventilator hospital-acquired pneumonia (nvHAP) are significant contributors to patient morbidity and mortality. Automated surveillance systems for these healthcare-associated infections have emerged as a potentially beneficial replacement for manual surveillance. This systematic review aims to synthesise the existing literature on the characteristics and performance of automated nvHAP and HAP surveillance systems.

METHODS

We conducted a systematic search of publications describing automated surveillance of nvHAP and HAP. Our inclusion criteria covered articles that described fully and semi-automated systems without limitations on patient demographics or healthcare settings. We detailed the algorithms in each study and reported the performance characteristics of automated systems that were validated against specific reference methods. Two published metrics were employed to assess the quality of the included studies.

RESULTS

Our review identified 12 eligible studies that collectively describe 24 distinct candidate definitions, 23 for fully automated systems and one for a semi-automated system. These systems were employed exclusively in high-income countries and the majority were published after 2018. The algorithms commonly included radiology, leukocyte counts, temperature, antibiotic administration, and microbiology results. Validated surveillance systems' performance varied, with sensitivities for fully automated systems ranging from 40 to 99%, specificities from 58 and 98%, and positive predictive values from 8 to 71%. Validation was often carried out on small, pre-selected patient populations.

CONCLUSIONS

Recent years have seen a steep increase in publications on automated surveillance systems for nvHAP and HAP, which increase efficiency and reduce manual workload. However, the performance of fully automated surveillance remains moderate when compared to manual surveillance. The considerable heterogeneity in candidate surveillance definitions and reference standards, as well as validation on small or pre-selected samples, limits the generalisability of the findings. Further research, involving larger and broader patient populations is required to better understand the performance and applicability of automated nvHAP surveillance.

Event-Based Clinical Finding Extraction from Radiology Reports with Pre-trained Language Model

Radiology reports contain a diverse and rich set of clinical abnormalities documented by radiologists during their interpretation of the images. Comprehensive semantic representations of radiological findings would enable a wide range of secondary use applications to support diagnosis, triage, outcomes prediction, and clinical research. In this paper, we present a new corpus of radiology reports annotated with clinical findings. Our annotation schema captures detailed representations of pathologic findings that are observable on imaging ("lesions") and other types of clinical problems ("medical problems"). The schema used an event-based representation to capture fine-grained details, including assertion, anatomy, characteristics, size, and count. Our gold standard corpus contained a total of 500 annotated computed tomography (CT) reports. We extracted triggers and argument entities using two state-of-the-art deep learning architectures, including BERT. We then predicted the linkages between trigger and argument entities (referred to as argument roles) using a BERT-based relation extraction model. We achieved the best extraction performance using a BERT model pre-trained on 3 million radiology reports from our institution: 90.9-93.4% F1 for finding triggers and 72.0-85.6% F1 for argument roles. To assess model generalizability, we used an external validation set randomly sampled from the MIMIC Chest X-ray (MIMIC-CXR) database. The extraction performance on this validation set was 95.6% for finding triggers and 79.1-89.7% for argument roles, demonstrating that the model generalized well to the cross-institutional data with a different imaging modality. We extracted the finding events from all the radiology reports in the MIMIC-CXR database and provided the extractions to the research community.

Electronically assisted surveillance systems of healthcare-associated infections: a systematic review

Background

Surveillance of healthcare-associated infections (HAI) is the basis of each infection control programme and, in case of acute care hospitals, should ideally include all hospital wards, medical specialties as well as all types of HAI. Traditional surveillance is labour intensive and electronically assisted surveillance systems (EASS) hold the promise to increase efficiency.

Objectives

To give insight in the performance characteristics of different approaches to EASS and the quality of the studies designed to evaluate them.

Methods

In this systematic review, online databases were searched and studies that compared an EASS with a traditional surveillance method were included. Two different indicators were extracted from each study, one regarding the quality of design (including reporting efficiency) and one based on the performance (e.g. specificity and sensitivity) of the EASS presented.

Results

A total of 78 studies were included. The majority of EASS (n = 72) consisted of an algorithm-based selection step followed by confirmatory assessment. The algorithms used different sets of variables. Only a minority (n = 7) of EASS were hospital-wide and designed to detect all types of HAI. Sensitivity of EASS was generally high (> 0.8), but specificity varied (0.37–1). Less than 20% (n = 14) of the studies presented data on the efficiency gains achieved.

Conclusions

Electronically assisted surveillance of HAI has yet to reach a mature stage and to be used routinely in healthcare settings. We recommend that future studies on the development and implementation of EASS of HAI focus on thorough validation, reproducibility, standardised datasets and detailed information on efficiency.

Automated screening of natural language in electronic health records for the diagnosis septic shock is feasible and outperforms an approach based on explicit administrative codes

PURPOSE

Identification of patients for epidemiologic research through administrative coding has important limitations. We investigated the feasibility of a search based on natural language processing (NLP) on the text sections of electronic health records for identification of patients with septic shock.

MATERIALS AND METHODS

Results of an explicit search strategy (using explicit concept retrieval) and a combined search strategy (using both explicit and implicit concept retrieval) were compared to hospital ICD-9 based administrative coding and to our department's own prospectively compiled infection database.

RESULTS

Of 8911 patients admitted to the medical or surgical ICU, 1023 (11.5%) suffered from septic shock according to the combined search strategy. This was significantly more than those identified by the explicit strategy (518, 5.8%), by hospital administrative coding (549, 5.8%) or by our own prospectively compiled database (609, 6.8%) (p < .001). Sensitivity and specificity of the automated combined search strategy were 72.7% (95%CI 69.0%-76.2%) and 93.0% (95%CI 92.4%-93.6%), compared to 56.0% (95%CI 52.0%-60.0%) and 97.5% (95%CI 97.1%-97.8%) for hospital administrative coding.

CONCLUSIONS

An automated search strategy based on a combination of explicit and implicit concept retrieval is feasible to screen electronic health records for septic shock and outperforms an administrative coding based explicit approach.

Development and Validation of a Natural Language Processing Tool to Identify Patients Treated for Pneumonia across VA Emergency Departments

BACKGROUND

Identifying pneumonia using diagnosis codes alone may be insufficient for research on clinical decision making. Natural language processing (NLP) may enable the inclusion of cases missed by diagnosis codes.

OBJECTIVES

This article (1) develops a NLP tool that identifies the clinical assertion of pneumonia from physician emergency department (ED) notes, and (2) compares classification methods using diagnosis codes versus NLP against a gold standard of manual chart review to identify patients initially treated for pneumonia.

METHODS

Among a national population of ED visits occurring between 2006 and 2012 across the Veterans Affairs health system, we extracted 811 physician documents containing search terms for pneumonia for training, and 100 random documents for validation. Two reviewers annotated span- and document-level classifications of the clinical assertion of pneumonia. An NLP tool using a support vector machine was trained on the enriched documents. We extracted diagnosis codes assigned in the ED and upon hospital discharge and calculated performance characteristics for diagnosis codes, NLP, and NLP plus diagnosis codes against manual review in training and validation sets.

RESULTS

Among the training documents, 51% contained clinical assertions of pneumonia; in the validation set, 9% were classified with pneumonia, of which 100% contained pneumonia search terms. After enriching with search terms, the NLP system alone demonstrated a recall/sensitivity of 0.72 (training) and 0.55 (validation), and a precision/positive predictive value (PPV) of 0.89 (training) and 0.71 (validation). ED-assigned diagnostic codes demonstrated lower recall/sensitivity (0.48 and 0.44) but higher precision/PPV (0.95 in training, 1.0 in validation); the NLP system identified more "possible-treated" cases than diagnostic coding. An approach combining NLP and ED-assigned diagnostic coding classification achieved the best performance (sensitivity 0.89 and PPV 0.80).

CONCLUSION

System-wide application of NLP to clinical text can increase capture of initial diagnostic hypotheses, an important inclusion when studying diagnosis and clinical decision-making under uncertainty.

Natural Language Processing in Radiology: A Systematic Review

Radiological reporting has generated large quantities of digital content within the electronic health record, which is potentially a valuable source of information for improving clinical care and supporting research. Although radiology reports are stored for communication and documentation of diagnostic imaging, harnessing their potential requires efficient and automated information extraction: they exist mainly as free-text clinical narrative, from which it is a major challenge to obtain structured data. Natural language processing (NLP) provides techniques that aid the conversion of text into a structured representation, and thus enables computers to derive meaning from human (ie, natural language) input. Used on radiology reports, NLP techniques enable automatic identification and extraction of information. By exploring the various purposes for their use, this review examines how radiology benefits from NLP. A systematic literature search identified 67 relevant publications describing NLP methods that support practical applications in radiology. This review takes a close look at the individual studies in terms of tasks (ie, the extracted information), the NLP methodology and tools used, and their application purpose and performance results. Additionally, limitations, future challenges, and requirements for advancing NLP in radiology will be discussed.

Patient walk detection in hospital room using Microsoft Kinect V2

This paper describes a system using Kinect sensor to detect patient walk automatically in a hospital room setting. The system is especially essential for the case when the patient is alone and the nursing staff is absent. The patient activities are represented by the features extracted from Kinect V2 skeletons. The analysis to the recognized walk could help us to better understand the health situation of the patient and the possible hospital acquired infection (HAI), and provide valuable information to healthcare givers for making a corresponding treatment decision and alteration. The Kinect V2 depth sensor provides the ground truth.

Development of trigger-based semi-automated surveillance of ventilator-associated pneumonia and central line-associated bloodstream infections in a Dutch intensive care

Background

Availability of a patient data management system (PDMS) has created the opportunity to develop trigger-based electronic surveillance systems (ESSs). The aim was to evaluate a semi-automated trigger-based ESS for the detection of ventilator-associated pneumonia (VAP) and central line-associated blood stream infections (CLABSIs) in the intensive care.

Methods

Prospective comparison of surveillance was based on a semi-automated ESS with and without trigger. Components of the VAP/CLABSI definition served as triggers. These included the use of VAP/CLABSI-related antibiotics, the presence of mechanical ventilation or an intravenous central line, and the presence of specific clinical symptoms. Triggers were automatically fired by the PDMS. Chest X-rays and microbiology culture results were checked only on patient days with a positive trigger signal from the ESS. In traditional screening, no triggers were used; therefore, chest X-rays and culture results had to be screened for all patient days of all included patients. Patients with pneumonia at admission were excluded.

Results

A total of 553 patients were screened for VAP and CLABSI. The incidence of VAP was 3.3/1,000 ventilation days (13 VAP/3,927 mechanical ventilation days), and the incidence of CLABSI was 1.7/1,000 central line days (24 CLABSI/13.887 central line days). For VAP, the trigger-based screening had a sensitivity of 92.3%, a specificity of 100%, and a negative predictive value of 99.8% compared to traditional screening of all patients. For CLABSI, sensitivity was 91.3%, specificity 100%, and negative predictive value 99.6%.

Conclusions

Pre-selection of patients to be checked for signs and symptoms of VAP and CLABSI by a computer-generated automated trigger system was time saving but slightly less accurate than conventional surveillance. However, this after-the-fact surveillance was mainly designed as a quality indicator over time rather than for precise determination of infection rates. Therefore, surveillance of VAP and CLABSI with a trigger-based ESS is feasible and effective.

Facilitating surveillance of pulmonary invasive mold diseases in patients with haematological malignancies by screening computed tomography reports using natural language processing

Purpose

Prospective surveillance of invasive mold diseases (IMDs) in haematology patients should be standard of care but is hampered by the absence of a reliable laboratory prompt and the difficulty of manual surveillance. We used a high throughput technology, natural language processing (NLP), to develop a classifier based on machine learning techniques to screen computed tomography (CT) reports supportive for IMDs.

Patients and Methods

We conducted a retrospective case-control study of CT reports from the clinical encounter and up to 12-weeks after, from a random subset of 79 of 270 case patients with 33 probable/proven IMDs by international definitions, and 68 of 257 uninfected-control patients identified from 3 tertiary haematology centres. The classifier was trained and tested on a reference standard of 449 physician annotated reports including a development subset (n = 366), from a total of 1880 reports, using 10-fold cross validation, comparing binary and probabilistic predictions to the reference standard to generate sensitivity, specificity and area under the receiver-operating-curve (ROC).

Results

For the development subset, sensitivity/specificity was 91% (95%CI 86% to 94%)/79% (95%CI 71% to 84%) and ROC area was 0.92 (95%CI 89% to 94%). Of 25 (5.6%) missed notifications, only 4 (0.9%) reports were regarded as clinically significant.

Conclusion

CT reports are a readily available and timely resource that may be exploited by NLP to facilitate continuous prospective IMD surveillance with translational benefits beyond surveillance alone.

Automated identification of pneumonia in chest radiograph reports in critically ill patients

BACKGROUND

Prior studies demonstrate the suitability of natural language processing (NLP) for identifying pneumonia in chest radiograph (CXR) reports, however, few evaluate this approach in intensive care unit (ICU) patients.

METHODS

From a total of 194,615 ICU reports, we empirically developed a lexicon to categorize pneumonia-relevant terms and uncertainty profiles. We encoded lexicon items into unique queries within an NLP software application and designed an algorithm to assign automated interpretations ('positive', 'possible', or 'negative') based on each report's query profile. We evaluated algorithm performance in a sample of 2,466 CXR reports interpreted by physician consensus and in two ICU patient subgroups including those admitted for pneumonia and for rheumatologic/endocrine diagnoses.

RESULTS

Most reports were deemed 'negative' (51.8%) by physician consensus. Many were 'possible' (41.7%); only 6.5% were 'positive' for pneumonia. The lexicon included 105 terms and uncertainty profiles that were encoded into 31 NLP queries. Queries identified 534,322 'hits' in the full sample, with 2.7 ± 2.6 'hits' per report. An algorithm, comprised of twenty rules and probability steps, assigned interpretations to reports based on query profiles. In the validation set, the algorithm had 92.7% sensitivity, 91.1% specificity, 93.3% positive predictive value, and 90.3% negative predictive value for differentiating 'negative' from 'positive'/'possible' reports. In the ICU subgroups, the algorithm also demonstrated good performance, misclassifying few reports (5.8%).

CONCLUSIONS

Many CXR reports in ICU patients demonstrate frank uncertainty regarding a pneumonia diagnosis. This electronic tool demonstrates promise for assigning automated interpretations to CXR reports by leveraging both terms and uncertainty profiles.

Using the electronic medical record to identify community-acquired pneumonia: toward a replicable automated strategy

BACKGROUND

Timely information about disease severity can be central to the detection and management of outbreaks of acute respiratory infections (ARI), including influenza. We asked if two resources: 1) free text, and 2) structured data from an electronic medical record (EMR) could complement each other to identify patients with pneumonia, an ARI severity landmark.

METHODS

A manual EMR review of 2747 outpatient ARI visits with associated chest imaging identified x-ray reports that could support the diagnosis of pneumonia (kappa score  = 0.88 (95% CI 0.82∶0.93)), along with attendant cases with Possible Pneumonia (adds either cough, sputum, fever/chills/night sweats, dyspnea or pleuritic chest pain) or with Pneumonia-in-Plan (adds pneumonia stated as a likely diagnosis by the provider). The x-ray reports served as a reference to develop a text classifier using machine-learning software that did not require custom coding. To identify pneumonia cases, the classifier was combined with EMR-based structured data and with text analyses aimed at ARI symptoms in clinical notes.

RESULTS

370 reference cases with Possible Pneumonia and 250 with Pneumonia-in-Plan were identified. The x-ray report text classifier increased the positive predictive value of otherwise identical EMR-based case-detection algorithms by 20-70%, while retaining sensitivities of 58-75%. These performance gains were independent of the case definitions and of whether patients were admitted to the hospital or sent home. Text analyses seeking ARI symptoms in clinical notes did not add further value.

CONCLUSION

Specialized software development is not required for automated text analyses to help identify pneumonia patients. These results begin to map an efficient, replicable strategy through which EMR data can be used to stratify ARI severity.

Advances in electronic surveillance for healthcare-associated infections in the 21st Century: a systematic review

BACKGROUND

Traditional methodologies for healthcare-associated infection (HCAI) surveillance can be resource intensive and time consuming. As a consequence, surveillance is often limited to specific organisms or conditions. Various electronic databases exist within the healthcare setting and may be utilized to perform HCAI surveillance.

AIM

To assess the utility of electronic surveillance systems for monitoring and detecting HCAI.

METHODS

A systematic review of published literature on surveillance of HCAI was performed. Databases were searched for studies published between January 2000 and December 2011. Search terms were divided into infection, surveillance and data management terms, and combined using Boolean operators. Studies were included for review if they demonstrated or proposed the use of electronic systems for HCAI surveillance.

FINDINGS

In total, 44 studies met the inclusion criteria. For the majority of studies, emphasis was on the linkage of electronic databases to provide automated methods for monitoring infections in specific clinical settings. Twenty-one studies assessed the performance of their method with traditional surveillance methodologies or a manual reference method. Where sensitivity and specificity were calculated, these varied depending on the organism or condition being surveyed and the data sources employed.

CONCLUSIONS

The implementation of electronic surveillance was found to be feasible in many settings, with several systems fully integrated into hospital information systems and routine surveillance practices. The results of this review suggest that electronic surveillance systems should be developed to maximize the efficacy of abundant electronic data sources existing within hospitals.

Bundling hand hygiene interventions and measurement to decrease health care-associated infections

Proper performance of hand hygiene at key moments during patient care is the most important means of preventing health care-associated infections (HAIs). With increasing awareness of the cost and societal impact caused by HAIs has come the realization that hand hygiene improvement initiatives are crucial to reducing the burden of HAIs. Multimodal strategies have emerged as the best approach to improving hand hygiene compliance. These strategies use a variety of intervention components intended to address obstacles to complying with good hand hygiene practices, and to reinforce behavioral change. Although research has substantiated the effectiveness of the multimodal design, challenges remain in promoting widespread adoption and implementation of a coordinated approach. This article reviews elements of a multimodal approach to improve hand hygiene and advocates the use of a "bundled" strategy. Eight key components of this bundle are proposed as a cohesive program to enable the deployment of synergistic, coordinated efforts to promote good hand hygiene practice. A consistent, bundled methodology implemented at multiple study centers would standardize processes and allow comparison of outcomes, validation of the methodology, and benchmarking. Most important, a bundled approach can lead to sustained infection reduction.

Pneumonia identification using statistical feature selection

OBJECTIVE

This paper describes a natural language processing system for the task of pneumonia identification. Based on the information extracted from the narrative reports associated with a patient, the task is to identify whether or not the patient is positive for pneumonia.

DESIGN

A binary classifier was employed to identify pneumonia from a dataset of multiple types of clinical notes created for 426 patients during their stay in the intensive care unit. For this purpose, three types of features were considered: (1) word n-grams, (2) Unified Medical Language System (UMLS) concepts, and (3) assertion values associated with pneumonia expressions. System performance was greatly increased by a feature selection approach which uses statistical significance testing to rank features based on their association with the two categories of pneumonia identification.

RESULTS

Besides testing our system on the entire cohort of 426 patients (unrestricted dataset), we also used a smaller subset of 236 patients (restricted dataset). The performance of the system was compared with the results of a baseline previously proposed for these two datasets. The best results achieved by the system (85.71 and 81.67 F1-measure) are significantly better than the baseline results (50.70 and 49.10 F1-measure) on the restricted and unrestricted datasets, respectively.

CONCLUSION

Using a statistical feature selection approach that allows the feature extractor to consider only the most informative features from the feature space significantly improves the performance over a baseline that uses all the features from the same feature space. Extracting the assertion value for pneumonia expressions further improves the system performance.

Using electronic patient records to discover disease correlations and stratify patient cohorts

Electronic patient records remain a rather unexplored, but potentially rich data source for discovering correlations between diseases. We describe a general approach for gathering phenotypic descriptions of patients from medical records in a systematic and non-cohort dependent manner. By extracting phenotype information from the free-text in such records we demonstrate that we can extend the information contained in the structured record data, and use it for producing fine-grained patient stratification and disease co-occurrence statistics. The approach uses a dictionary based on the International Classification of Disease ontology and is therefore in principle language independent. As a use case we show how records from a Danish psychiatric hospital lead to the identification of disease correlations, which subsequently can be mapped to systems biology frameworks.

Text mining and information extraction can be seen as the challenge of converting information hidden in text into manageable data. We have used text mining to automatically extract clinically relevant terms from 5543 psychiatric patient records and map these to disease codes in the International Classification of Disease ontology (ICD10). Mined codes were supplemented by existing coded data. For each patient we constructed a phenotypic profile of associated ICD10 codes. This allowed us to cluster patients together based on the similarity of their profiles. The result is a patient stratification based on more complete profiles than the primary diagnosis, which is typically used. Similarly we investigated comorbidities by looking for pairs of disease codes cooccuring in patients more often than expected. Our high ranking pairs were manually curated by a medical doctor who flagged 93 candidates as interesting. For a number of these we were able to find genes/proteins known to be associated with the diseases using the OMIM database. The disease-associated proteins allowed us to construct protein networks suspected to be involved in each of the phenotypes. Shared proteins between two associated diseases might provide insight to the disease comorbidity.

Automated detection of nosocomial infections: evaluation of different strategies in an intensive care unit 2000-2006

The aim of this study was to evaluate seven different strategies for the automated detection of nosocomial infections (NIs) in an intensive care unit (ICU) by using different hospital information systems: microbiology database, antibiotic prescriptions, medico-administrative database, and textual hospital discharge summaries. The study involved 1,499 patients admitted to an ICU of the University Hospital of Lyon (France) between 2000 and 2006. The data were extracted from the microbiology laboratory information system, the clinical information system on the ward and the medico-administrative database. Different algorithms and strategies were developed, using these data sources individually or in combination. The performances of each strategy were assessed by comparing the results with the ward data collected as a national standardised surveillance protocol, adapted from the National Nosocomial Infections Surveillance system as the gold standard. From 1,499 patients, 282 NIs were reported. The strategy with the best sensitivity for detecting these infections using an automated method was the combination of antibiotic prescription or microbiology, with a sensitivity of 99.3% [95% confidence interval (CI): 98.2-100] and a specificity of 56.8% (95% CI: 54.0-59.6). Automated methods of NI detection represent an alternative to traditional monitoring methods. Further study involving more ICUs should be performed before national recommendations can be established.

A review of human factors principles for the design and implementation of medication safety alerts in clinical information systems

The objective of this review is to describe the implementation of human factors principles for the design of alerts in clinical information systems. First, we conduct a review of alarm systems to identify human factors principles that are employed in the design and implementation of alerts. Second, we review the medical informatics literature to provide examples of the implementation of human factors principles in current clinical information systems using alerts to provide medication decision support. Last, we suggest actionable recommendations for delivering effective clinical decision support using alerts. A review of studies from the medical informatics literature suggests that many basic human factors principles are not followed, possibly contributing to the lack of acceptance of alerts in clinical information systems. We evaluate the limitations of current alerting philosophies and provide recommendations for improving acceptance of alerts by incorporating human factors principles in their design.

Developing information technology for infection prevention surveillance

The potential to automate at least part of the surveillance process for health care-associated infections was seen as soon as hospitals began to implement computer systems. Progress toward automated surveillance has been ongoing for the last several decades. But as more information becomes available electronically in the healthcare setting, the promise of electronic surveillance for healthcare-associated infections has become closer to reality. Although true fully automated surveillance is not here yet, significant progress is being made at a number of centers for electronic surveillance of central catheter-associated bloodstream infections, ventilator-associated pneumonia, and other healthcare-associated infections. We review the progress that has been made in this area and issues that need to be addressed as surveillance systems are implemented, as well as promising areas for future development.

Practical experience with the maintenance and auditing of a large medical ontology

The Medical Entities Dictionary (MED) has served as a unified terminology at New York Presbyterian Hospital and Columbia University for more than 20 years. It was initially created to allow the clinical data from the disparate information systems (e.g., radiology, pharmacy, and multiple laboratories, etc.) to be uniquely codified for storage in a single data repository, and functions as a real time terminology server for clinical applications and decision support tools. Being conceived as a knowledge base, the MED incorporates relationships among local terms, between local terms and external standards, and additional knowledge about terms in a semantic network structure. Over the past two decades, we have sought to develop methods to maintain, audit and improve the content of the MED, such that it remains true to its original design goals. This has resulted in a complex, multi-faceted process, with both manual and automated components. In this paper, we describe this process, with examples of its effectiveness. We believe that our process provides lessons for others who seek to maintain complex, concept-oriented controlled terminologies.

Strategies to prevent ventilator-associated pneumonia in acute care hospitals

Previously published guidelines are available that provide comprehensive recommendations for detecting and preventing healthcare-associated infections. The intent of this document is to highlight practical recommendations in a concise format designed to assist acute care hospitals in implementing and prioritizing their ventilator-associated pneumonia (VAP) prevention efforts. Refer to the Society for Healthcare Epidemiology of America/Infectious Diseases Society of America “Compendium of Strategies to Prevent Healthcare-Associated Infections” Executive Summary and Introduction and accompanying editorial for additional discussion.

1. Occurrence of VAP in acute care facilities.

a. VAP is one of the most common infections acquired by adults and children in intensive care units (ICUs).

i. In early studies, it was reported that 10%-20% of patients undergoing ventilation developed VAP. More-recent publications report rates of VAP that range from 1 to 4 cases per 1,000 ventilator-days, but rates may exceed 10 cases per 1,000 ventilator-days in some neonatal and surgical patient populations. The results of recent quality improvement initiatives, however, suggest that many cases of VAP might be prevented by careful attention to the process of care.

2. Outcomes associated with VAP

a. VAP is a cause of significant patient morbidity and mortality, increased utilization of healthcare resources, and excess cost.

i. The mortality attributable to VAP may exceed 10%.

ii. Patients with VAP require prolonged periods of mechanical ventilation, extended hospitalizations, excess use of antimicrobial medications, and increased direct medical costs.

Automatic segmentation of clinical texts

Clinical narratives, such as radiology and pathology reports, are commonly available in electronic form. However, they are also commonly entered and stored as free text. Knowledge of the structure of clinical narratives is necessary for enhancing the productivity of healthcare departments and facilitating research. This study attempts to automatically segment medical reports into semantic sections. Our goal is to develop a robust and scalable medical report segmentation system requiring minimum user input for efficient retrieval and extraction of information from free-text clinical narratives. Hand-crafted rules were used to automatically identify a high-confidence training set. This automatically created training dataset was later used to develop metrics and an algorithm that determines the semantic structure of the medical reports. A word-vector cosine similarity metric combined with several heuristics was used to classify each report sentence into one of several pre-defined semantic sections. This baseline algorithm achieved 79% accuracy. A Support Vector Machine (SVM) classifier trained on additional formatting and contextual features was able to achieve 90% accuracy. Plans for future work include developing a configurable system that could accommodate various medical report formatting and content standards.

Validity of electronic surveillance systems: a systematic review

Electronic surveillance that utilises information held in databases is more efficient than conventional infection surveillance methods. Validity is not well-defined, however. We systematically reviewed studies comparing the utility of electronic and conventional surveillance methods. Publications were identified using Medline (1980-2007) and bibliographic review. The sensitivity and specificity of electronic compared with conventional surveillance was reported. Twenty-four studies were included. Six studies reported that nosocomial infections could be detected utilising microbiology data alone with good overall sensitivity (range: 63-91%) and excellent specificity (range: 87 to >99%). Two studies used three laboratory-based algorithms for the detection of infection outbreaks yielding variable utility measures (sensitivity, range: 43-91%; specificity, range: 67-86%). Seven studies using only administrative data including discharge coding (International Classification of Diseases, 9th edn, Clinical Modification) and pharmacy data claimed databases had good sensitivity (range: 59-96%) and excellent specificity (range: 95 to >99%) in detecting nosocomial infections. Six studies combined both laboratory and administrative data for a range of infections, and overall had higher sensitivity (range: 71-94%) but lower specificity (range: 47 to >99%) than with use of either alone. Three studies evaluated community-acquired infections with variable results. Electronic surveillance has moderate to excellent utility compared with conventional methods for nosocomial infections. Future studies are needed to refine electronic algorithms further, especially with community-onset infections.

Economics of infection control surveillance technology: cost-effective or just cost?

BACKGROUND

Previous studies have suggested that informatics tools, such as automated alert and decision support systems, may increase the efficiency and quality of infection control surveillance. However, little is known about the cost-effectiveness of these tools.

METHODS

We focus on 2 types of economic analyses that have utility in assessing infection control interventions (cost-effectiveness analysis and business-case analysis) and review the available literature on the economics of computerized infection control surveillance systems.

RESULTS

Previous studies on the effectiveness of computerized infection control surveillance have been limited to assessments of whether these tools increase the sensitivity and specificity of surveillance over traditional methods. Furthermore, we identified only 2 studies that assessed the costs associated with computerized infection control surveillance. Thus, it remains unknown whether computerized infection control surveillance systems are cost-effective and whether use of these systems improves patient outcomes.

CONCLUSION

The existing data are insufficient to allow for a summary conclusion on the cost-effectiveness of infection control surveillance technology. All future studies of computerized infection control surveillance systems should aim to collect outcomes and economic data to inform decision making and assist hospitals with completing business-cases analyses.

Nosocomial infection in small for gestational age newborns with birth weight <1500 g: a multicentre analysis

OBJECTIVE

To investigate whether preterm newborns who are small for gestational age are at increased risk of nosocomial infections and necrotising enterocolitis.

DESIGN, SETTING AND SUBJECTS

The German national surveillance system for nosocomial infection in very low birthweight infants uses the US Centers for Disease Control and Prevention criteria. 2918 newborns (24-28 weeks), born between 2000 and 2004, were selected after application of predefined inclusion criteria to ensure similar proportions of small and appropriate weight for gestational age newborns across gestational age groups.

MAIN OUTCOME MEASURES

The outcome criterion was at least one episode of nosocomial sepsis, pneumonia or necrotising enterocolitis. Adjusted odds ratios and corresponding 95% CIs were calculated based on general estimating equation models.

RESULTS

The study population consisted of 13% (n = 392) small and 87% (n = 2526) appropriate weight for gestational age infants. 33% (n = 950) of the infants experienced at least one episode of sepsis: 42% (n = 163) of small and 31% (n = 787) of appropriate weight for gestational age newborns (adjusted OR 1.41, 95% CI 1.05 to 1.89). Pneumonia was diagnosed in 6% (n = 171) of infants: 8.4% (n = 33) of small and 5.5% (n = 138) of appropriate weight for gestational age newborns (adjusted OR 1.57, 95% CI 1.19 to 5.57). Necrotising enterocolitis was documented in 5.2% (n = 152) of infants: 7.1% (n = 28) of small and 4.9% of (n = 124) appropriate weight for gestational age newborns (adjusted OR 1.20, 95% confidence interval 0.75 to 1.94).

CONCLUSIONS

Growth-retarded preterm infants seem to be at increased risk of nosocomial infection, irrespective of the responsible pathogen. Future immunological research should elucidate potential causal associations.

Ventilator-associated pneumonia in neonatal and pediatric intensive care unit patients

Ventilator-associated pneumonia (VAP) is the second most common hospital-acquired infection among pediatric intensive care unit (ICU) patients. Empiric therapy for VAP accounts for approximately 50% of antibiotic use in pediatric ICUs. VAP is associated with an excess of 3 days of mechanical ventilation among pediatric cardiothoracic surgery patients. The attributable mortality and excess length of ICU stay for patients with VAP have not been defined in matched case control studies. VAP is associated with an estimated $30,000 in attributable cost. Surveillance for VAP is complex and usually performed using clinical definitions established by the CDC. Invasive testing via bronchoalveolar lavage increases the sensitivity and specificity of the diagnosis. The pathogenesis in children is poorly understood, but several prospective cohort studies suggest that aspiration and immunodeficiency are risk factors. Educational interventions and efforts to improve adherence to hand hygiene for children have been associated with decreased VAP rates. Studies of antibiotic cycling in pediatric patients have not consistently shown this measure to prevent colonization with multidrug-resistant gram-negative rods. More consistent and precise approaches to the diagnosis of pediatric VAP are needed to better define the attributable morbidity and mortality, pathophysiology, and appropriate interventions to prevent this disease.