Surveillance has been described as an essential component
of an effective infection prevention and control programme,
and has been shown to lead to a decrease in infection rates.1e3
However, 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. In 1993, Glenister
et al. concluded that selective surveillance methods for HCAI
are more appropriate in the healthcare setting than hospitalwide
continuous surveillance, which are too labour intensive.
4 These conclusions were made based on the use of manual
methods for the collection of data, where electronic data
sources were limited. Whole-hospital surveillance only tends to
be performed within point prevalence surveys, which are
extremely resource intensive. The results of point prevalence
surveys can help identify issues requiring further investigation,
but they cannot provide real-time information as the data
collation, analysis and feedback process is extensive. In recent
years, however, the collection and storage of healthcare information
in electronic formats has become ubiquitous, and
various electronic databases exist within the healthcare setting
that can be used to perform continuous HCAI surveillance.
In addition, surveillance schemes often rely on the application
of complex definitions which vary between countries.5
These definitions can be subjective and therefore have the
potential to introduce variation across institutions participating
in national surveillance schemes. With increasing
emphasis on benchmarking and public reporting of surveillance
data,6 surveillance methods need to be efficient, reliable and
transferable. A systematic review was therefore undertaken to
assess the utility of electronic surveillance for HCAI, and
determine the performance and feasibility of the methods and
systems described.
……………………………………………
An intentionally broad systematic search was undertaken to
locate articles published in journals across several disciplines.
The search aimed to capture articles proposing novel electronic
surveillance systems, as well as studies demonstrating
the utility of electronic surveillance within healthcare settings.
Databases were searched for articles published between
January 2000 and December 2011. MEDLINE (via PubMed), the
Cochrane Library (specifically the Cochrane Methodology Register
and Health Technology Assessment Database), Ovid,
EMBASE, Web of Science, Scopus, JSTOR and Wiley Online Library
were searched for relevant articles within this time
period. BIOSIS Previews was searched for conference and
meeting abstracts published between 1998 and 2008 due to
access limitations. As stated in the Cochrane Handbook of
Systematic Reviews, filters, language restrictions and format
restrictions were not introduced at this stage in order to achieve
high sensitivity.7 Manual searches were performed by
exploring the reference lists of all evaluated articles.
A combination of Medical Subject Heading searches and
free-text searches was performed using Boolean operators.
Search terms were divided into three categories: (i) infection
terms; (ii) surveillance terms; and (iii) data management
terms. In order to achieve a comprehensive overarching
search, a wide range of free-text terms were selected to
include synonyms, related terms and variant spellings. Truncation
and wildcard features were also incorporated into
search terms. Although standard terms were used across all
databases, search strategies were customized for each database
to account for the differences between the interfaces and
capabilities of each database.
Inclusion and exclusion criteria
Studies were included if they either presented a working
electronic surveillance system (ESS) for detecting or monitoring
HCAI, or if they proposed/described an electronic system
for HCAI surveillance. Retrospective and prospective
cohort studies, surveillance studies, validation studies, observational
studies and system descriptions were included if they
were published in English. Outbreak detection systems were
excluded as these will be reviewed as part of a different study.
A random sample of papers included for full-text evaluation
were assessed against the inclusion criteria by a second
reviewer to validate that the first reviewer had applied the
criteria successfully to all studies evaluated.
Quality assessment
The Strengthening the Reporting of Observational Studies in
Epidemiology8,9 checklist for cohort studies and the Updated
Guidelines for Evaluating Public Health Surveillance Systems10
were merged to create a unique quality criteria tool to assess
included studies.
Data extraction
All search results were imported into reference manager
software where duplicates were identified and removed. The
titles and abstracts of all retrieved articles were screened to
assess their relevance. After all irrelevant studies were identified
and removed, a list of potentially relevant results
remained and full-text retrieval was attempted. Full-text
evaluation was performed to determine if each study satisfied
the inclusion criteria.
Information was extracted and collated in data collection
forms. For all included studies, the data sources used for surveillance,
organisms or conditions under surveillance, and the
clinical specialty surveyed were recorded. Sensitivity, specificity,
positive predictive value (PPV), negative predictive
value (NPV) and the kappa statistic of agreement were recorded
if reported within validation studies.
Results
The initial search returned 44,765 results. Figure 1 summarizes
the process undertaken to identify all relevant articles.
No studies were excluded following quality assessment.
Although some papers did score low in terms of quality, these
tended to be conference proceedings papers rather than articles
published in scientific journals. Forty-four articles were
therefore included in the final review.
Tables IeIII summarize each article included in this review.
Thirty-seven articles presented methods utilizing several data
sources, including microbiological data. Of the 44 articles
included for review, 21 were validation studies. The validation
studies compared the performance of the reported electronic
surveillance with traditional surveillance methodologies or a
manual referencemethod. Table IV summarizes the performance
of each electronic method compared with the authors’ chosen
reference method. The performance of electronic surveillance
methods varied across infections. Consistently high sensitivity
results were observed across studies focusing on bloodstream
infection (BSI, 72e100%), although performance in terms of
specificity varied greatly (37e100%). Validation studies on urinary
tract infections (UTI) provided similar results, with sensitivities
ranging from 86% to 100% and specificities ranging from 59% to
100%. The reported sensitivities for ESSs detecting cases of surgical
site infection (SSI) ranged from 60% to 98%, with one study
reportinganNPVof 100%,andsensitivity and specificity results for
systems designed to detect cases of pneumonia ranged from 71%
to 99% and 61% to 100%, respectively.
The majority of studies involved the development of ‘inhouse’
surveillance systems rather than employing commercially
available software. Dao et al. suggested that significant
costs associated with purchasing commercial software packages
could be avoided if existing databases within the hospital
were exploited instead.11 However, a number of studies did
involve the development or use of commercial software packages
for electronic surveillance of HCAI, and these are indicated
in Tables I and III. Systems and methods described in the
literature have been categorized and are described in detail
below.
Incorporation of standard definitions for electronic
surveillance
Several authors incorporated established HCAI surveillance
definitions into their systems.12e16 Choudhuri et al. created an
electronic surveillance tool based on adapted versions of the
Centers for Disease Control and Prevention’s National Healthcare
Safety Network (CDC-NHSN) definition for catheterassociated
UTI.12 Routinely collected clinical and laboratory
data were extracted from the hospital information system, and
results were compared with traditional manual review of
electronic records by an infection prevention and control
specialist. The authors found that the electronic system
allowed more efficient surveillance than was achieved with
manual chart review. Klompas et al. employed a similar
approach, which mirrored the CDC definitions, although
quantitative thresholds were developed to replace qualitative
criteria.13 The system presented by Adlassnig et al. involved
the use of rules based on established HCAI definitions and
several programmes which allowed the monitoring of organisms
exhibiting specific resistance patterns, the identification
of potential cross-infection incidents and the detection of
increasing rates of infection.15
Data warehouses for electronic surveillance
The creation of data warehouses for the surveillance of HCAI
has been presented by several authors.17e22 In the system
developed by Bellini et al., data were processed through a
multi-step algorithm to exclude contaminated blood cultures,
distinguish between community and nosocomial BSI, and classify
BSI as catheter associated or non-catheter associated.17
The results of the automated surveillance were highly
concordant with the manual surveillance method. The authors
did acknowledge that the automated surveillance system
differed from the CDC surveillance criteria, as clinical details
required to satisfy the definitions were not available in the
hospital information system, yet the sensitivity and specificity
for the detection of BSI were high (Table IV).
Wisniewski et al. developed a relational clinical data
warehouse using existing data for the
เฝ้าระวังมีการอธิบายเป็นส่วนประกอบสำคัญของการป้องกันการติดเชื้อที่มีประสิทธิภาพและการควบคุมโครงการและจะนำไปสู่การลดลงติดเชื้อ rates.1e3อย่างไรก็ตาม วิธีการดั้งเดิมสำหรับดูแลสุขภาพเกี่ยวข้องกับเฝ้าระวังการติดเชื้อ (HCAI) สามารถเร่งรัดทรัพยากร และใช้เวลานาน ผล เฝ้าระวังเป็นจำกัดเฉพาะสิ่งมีชีวิตหรือสภาพ ในปี 1993, Glenisteral. ร้อยเอ็ดสรุปการเฝ้าระวังที่ใช้วิธีการ HCAIเหมาะมากในการดูแลสุขภาพมากกว่า hospitalwideอย่างต่อเนื่องเฝ้าระวัง ซึ่งมากเกินไปแรงงาน4 บทสรุปของเหล่านี้ถูกสร้างขึ้นตามคู่มือการใช้วิธีการเก็บรวบรวมข้อมูล ซึ่งข้อมูลอิเล็กทรอนิกส์แหล่งที่ถูกจำกัด โรงพยาบาลทั้งหมดเฝ้าระวังเท่านั้นมีแนวโน้มทำงานในจุดชุกสำรวจ ซึ่งเป็นมากทรัพยากรแบบเร่งรัด ผลลัพธ์ของจุดชุกสำรวจสามารถช่วยระบุปัญหาที่ต้องการตรวจสอบ เพิ่มเติมแต่พวกเขาไม่สามารถให้ข้อมูลเวลาจริงเป็นข้อมูลกระบวนการเปรียบเทียบ วิเคราะห์ และข้อเสนอแนะเป็นอย่างละเอียด ในล่าสุดปี อย่างไรก็ตาม การเก็บรวบรวมและจัดเก็บข้อมูลสุขภาพในรูปแบบอิเล็กทรอนิกส์ได้กลายเป็นแพร่หลาย และฐานข้อมูลอิเล็กทรอนิกส์ต่าง ๆ ที่มีอยู่ภายในค่าดูแลสุขภาพที่สามารถใช้เพื่อดำเนินการเฝ้าระวังอย่างต่อเนื่อง HCAIนอกจากนี้ แผนงานรักษาความปลอดภัยมักจะพึ่งแอพลิเคชันของข้อกำหนดที่ซับซ้อนซึ่งแตกต่างกันระหว่าง countries.5These definitions can be subjective and therefore have thepotential to introduce variation across institutions participatingin national surveillance schemes. With increasingemphasis on benchmarking and public reporting of surveillancedata,6 surveillance methods need to be efficient, reliable andtransferable. A systematic review was therefore undertaken toassess the utility of electronic surveillance for HCAI, anddetermine the performance and feasibility of the methods andsystems described.……………………………………………An intentionally broad systematic search was undertaken tolocate articles published in journals across several disciplines.The search aimed to capture articles proposing novel electronicsurveillance systems, as well as studies demonstratingthe utility of electronic surveillance within healthcare settings.Databases were searched for articles published betweenJanuary 2000 and December 2011. MEDLINE (via PubMed), theCochrane Library (specifically the Cochrane Methodology Registerand Health Technology Assessment Database), Ovid,EMBASE, Web of Science, Scopus, JSTOR and Wiley Online Librarywere searched for relevant articles within this timeperiod. BIOSIS Previews was searched for conference andmeeting abstracts published between 1998 and 2008 due toaccess limitations. As stated in the Cochrane Handbook ofSystematic Reviews, filters, language restrictions and formatrestrictions were not introduced at this stage in order to achieve
high sensitivity.7 Manual searches were performed by
exploring the reference lists of all evaluated articles.
A combination of Medical Subject Heading searches and
free-text searches was performed using Boolean operators.
Search terms were divided into three categories: (i) infection
terms; (ii) surveillance terms; and (iii) data management
terms. In order to achieve a comprehensive overarching
search, a wide range of free-text terms were selected to
include synonyms, related terms and variant spellings. Truncation
and wildcard features were also incorporated into
search terms. Although standard terms were used across all
databases, search strategies were customized for each database
to account for the differences between the interfaces and
capabilities of each database.
Inclusion and exclusion criteria
Studies were included if they either presented a working
electronic surveillance system (ESS) for detecting or monitoring
HCAI, or if they proposed/described an electronic system
for HCAI surveillance. Retrospective and prospective
cohort studies, surveillance studies, validation studies, observational
studies and system descriptions were included if they
were published in English. Outbreak detection systems were
excluded as these will be reviewed as part of a different study.
A random sample of papers included for full-text evaluation
were assessed against the inclusion criteria by a second
reviewer to validate that the first reviewer had applied the
criteria successfully to all studies evaluated.
Quality assessment
The Strengthening the Reporting of Observational Studies in
Epidemiology8,9 checklist for cohort studies and the Updated
Guidelines for Evaluating Public Health Surveillance Systems10
were merged to create a unique quality criteria tool to assess
included studies.
Data extraction
All search results were imported into reference manager
software where duplicates were identified and removed. The
titles and abstracts of all retrieved articles were screened to
assess their relevance. After all irrelevant studies were identified
and removed, a list of potentially relevant results
remained and full-text retrieval was attempted. Full-text
evaluation was performed to determine if each study satisfied
the inclusion criteria.
Information was extracted and collated in data collection
forms. For all included studies, the data sources used for surveillance,
organisms or conditions under surveillance, and the
clinical specialty surveyed were recorded. Sensitivity, specificity,
positive predictive value (PPV), negative predictive
value (NPV) and the kappa statistic of agreement were recorded
if reported within validation studies.
Results
The initial search returned 44,765 results. Figure 1 summarizes
the process undertaken to identify all relevant articles.
No studies were excluded following quality assessment.
Although some papers did score low in terms of quality, these
tended to be conference proceedings papers rather than articles
published in scientific journals. Forty-four articles were
therefore included in the final review.
Tables IeIII summarize each article included in this review.
Thirty-seven articles presented methods utilizing several data
sources, including microbiological data. Of the 44 articles
included for review, 21 were validation studies. The validation
studies compared the performance of the reported electronic
surveillance with traditional surveillance methodologies or a
manual referencemethod. Table IV summarizes the performance
of each electronic method compared with the authors’ chosen
reference method. The performance of electronic surveillance
methods varied across infections. Consistently high sensitivity
results were observed across studies focusing on bloodstream
infection (BSI, 72e100%), although performance in terms of
specificity varied greatly (37e100%). Validation studies on urinary
tract infections (UTI) provided similar results, with sensitivities
ranging from 86% to 100% and specificities ranging from 59% to
100%. The reported sensitivities for ESSs detecting cases of surgical
site infection (SSI) ranged from 60% to 98%, with one study
reportinganNPVof 100%,andsensitivity and specificity results for
systems designed to detect cases of pneumonia ranged from 71%
to 99% and 61% to 100%, respectively.
The majority of studies involved the development of ‘inhouse’
surveillance systems rather than employing commercially
available software. Dao et al. suggested that significant
costs associated with purchasing commercial software packages
could be avoided if existing databases within the hospital
were exploited instead.11 However, a number of studies did
involve the development or use of commercial software packages
for electronic surveillance of HCAI, and these are indicated
in Tables I and III. Systems and methods described in the
literature have been categorized and are described in detail
below.
Incorporation of standard definitions for electronic
surveillance
Several authors incorporated established HCAI surveillance
definitions into their systems.12e16 Choudhuri et al. created an
electronic surveillance tool based on adapted versions of the
Centers for Disease Control and Prevention’s National Healthcare
Safety Network (CDC-NHSN) definition for catheterassociated
UTI.12 Routinely collected clinical and laboratory
data were extracted from the hospital information system, and
results were compared with traditional manual review of
electronic records by an infection prevention and control
specialist. The authors found that the electronic system
allowed more efficient surveillance than was achieved with
manual chart review. Klompas et al. employed a similar
approach, which mirrored the CDC definitions, although
quantitative thresholds were developed to replace qualitative
criteria.13 The system presented by Adlassnig et al. involved
the use of rules based on established HCAI definitions and
several programmes which allowed the monitoring of organisms
exhibiting specific resistance patterns, the identification
of potential cross-infection incidents and the detection of
increasing rates of infection.15
Data warehouses for electronic surveillance
The creation of data warehouses for the surveillance of HCAI
has been presented by several authors.17e22 In the system
developed by Bellini et al., data were processed through a
multi-step algorithm to exclude contaminated blood cultures,
distinguish between community and nosocomial BSI, and classify
BSI as catheter associated or non-catheter associated.17
The results of the automated surveillance were highly
concordant with the manual surveillance method. The authors
did acknowledge that the automated surveillance system
differed from the CDC surveillance criteria, as clinical details
required to satisfy the definitions were not available in the
hospital information system, yet the sensitivity and specificity
for the detection of BSI were high (Table IV).
Wisniewski et al. developed a relational clinical data
warehouse using existing data for the
การแปล กรุณารอสักครู่..
