Knowledge-based monitoring of hospital-acquired infections in adult intensive care patients

Klaus-Peter Adlassnig1, Alexander Blacky2, Walter Koller2

1 Section on Medical Expert and Knowledge-Based SystemsMedical University of ViennaSpitalgasse 23A-1090 Vienna, Austria www.meduniwien.ac.at/mes

2

Division of Hospital HygieneClinical Institute for Hygiene and Medical MicrobiologyMedical University of ViennaWaehringer

Guertel

18–20A-1090 Vienna, Austria

Einführung in Medizinische Informatik, WS 2008/09, 5 November 2008

Computers in clinical medicine—steps of natural progression

•

step 1: patient administration–

admission, transfer, discharge, and billing of medical services

•

step 2: documentation of patients’

medical data–

electronic health record: life-long, multimedia

•

step 3: patient data retrieval and analysis–

medical research databases, medical studies–

quality assurance in the medical institution

•

step 4: knowledge-based software systems for clinical decision support–

safety net, quality assurance and improvement: …

for the individual patient … and the physician …

and the medical institution

•

studies in Colorado and Utah and in New York (1997)

– errors in the delivery of health care leading to the death of as many as 98,000 US citizens annually

•

causes of errors

– error or delay in diagnosis

– failure to employ indicated tests

– use of outmoded tests or therapy

– failure to act on results of testing or monitoring

– error in the performance of a test, procedure, or operation

– error in administering the treatment

– error in the dose or method of using a drug

– avoidable delay in treatment or in responding to an abnormal test

– inappropriate (not indicated) care

– failure of communication

– equipment failure

•

prevention of errors

– we must systematically design safety into processes of care

errors

prevention

Nosocomial, or hospital-acquired, infections

ESBL -

extended spectrum beta-lactamase

VRE -

vancomycin-resistant

enterococcus

MDR-TB

-

multidrug-resistant

tuberculosis

increaseddisposition by low immunity

MRSA -

methicillin-resistant

Staphylococcus aureus

exposure to pathogens

entry sites

Potential for reducing the rate of hospital-acquired infections

hospital-acquired (nosocomial) infections:

% reduction:

∗

wound infections

35∗

urinary tract infections

31 ∗

pneumonias

28 ∗

bloodstream infections

35

through continuous surveillance

Haley, R.W., Culver, D.H., White, J.W., Morgan, W. M., Emori, T.G., Munn, V.P., and Hooton, T.M. (1985) The efficacy of infection surveillance and control programs in preventing nosocomial infections in US hospitals. American Journal of Epidemiology 121(2), 182–205. (SENIC study)

Moni/Surveillance-ICU

knowledge-based identification and monitoring of nosocomial infections

patient-specific alerts

infection control

natural-language definitions of nosocomial

infections

Fuzzy theories

Artificial intelligence

Monitoring

of nosocomial infections

knowledge-based systems

fuzzy sets and logic

ICUICU

microbiology

cockpit surveillance remote

clinical data

Medicine

data on microorganisms

cockpit surveillance at ward ICU

Bloodstream infection with clinical signs and growth of same skin contaminant from two separate blood samples

BSI-A2

1⇐

clinical_signs_of_BSI (t-1d, t, t+1d)∧

same_skin_contaminant_from_two_separate_blood_samples

Decomposition—clinical signs

clinical_signs_of_BSI (t-1d, t, t+1d)[yesterday, today, tomorrow]

=fever (t-1d)

∨hypotension (t-1d)

∨clinical_signs_of_BSI (t-1d) = leucopenia (t-1d)

∨leucocytosis (t-1d)

∨CRP increased (t-1d)

∨fever (t)

∨hypotension (t)

∨clinical_signs_of_BSI (t) = leucopenia (t)

∨leucocytosis (t)

∨CRP increased (t)

∨fever (t+1d)

∨hypotension (t+1d)

∨clinical_signs_of_BSI (t+1d) = leucopenia (t+1d)

∨leucocytosis (t+1d)

∨CRP increased (t+1d)

fever (t-1d) ⇐

...

body temperature ↑

fever (t) ⇐ ∨

thermoregulation applied

fever (t+1d) ⇐

...

Clinical signs—fever

data import

intensive care unit

maximum value of the day

e.g., 38.5 °C°C

1

037 37.5 38 38.5

…

Crisp sets vs. fuzzy sets

yes/no decision

gradual transition

age

1

χY young

0 threshold

U = [0, 120]Y ⊆

U with Y = {(µY (x)/

x)⏐x ∈

U}µY : U →

[0, 1]1

1 + (0.04 x)2 ∀

x ∈

U0 age

1

μY young

threshold0

U = [0, 120]Y ⊆

U with Y = {(χY (x)/

x)⏐x ∈

U}χ Y : U →

{0, 1}

χ Y (x) = ∀

x ∈

U0

x > threshold1 x ≤ threshold

1 x ≤ threshold

x > thresholdµY (x) =

0

⎩⎨⎧

⎩⎨⎧

CRP increased (t-1d) ⇐

...

CRP increased (t) ⇐

CRP increased (t+1d) ⇐

...

Clinical signs—CRP increased

data import

intensive care unit

maximum value of the day

e.g., 5 mg/dlmg/dl

01 6

CRP

0.8

1

5

Decomposition—skin contaminant

first blood culture

- coagulase-negative staphylococci

- Micrococcus sp.

- Propionibacterium acnes

- Bacillus sp.

- Corynebacterium sp.

same_skin_contaminant_from_two_separate_blood_samples

second blood culture

- coagulase-negative staphylococci

- Micrococcus sp.

- Propionibacterium acnes

- Bacillus sp.

- Corynebacterium sp.

data import

microbiology∧

(within 48 hours)⇐

Cockpit surveillance at the infection control unit: Three criteria- based definitions in two patients are completely fulfilled (100%), backtracking of the logical chain of reasoning is provided …

… until the detailed medical data are reached

Data sources and integration

HIS: hospital information system (here: HIS of the City of Vienna)PDMS: patient data management systems (here: CareVue by Philips)CDA: clinical data archiveISM: information support martLIS: laboratory information system of the microbiology (here: HIS of the City of Vienna)HIS DB: relational data base of medical data

LIS

PDMSs

CDA

ISMHIS

DBHIS

…

Moni

clinical data

microbiological data

CareVueCareVue

CareVueCareVue

administrative data+

HIS

Processing layers

NI definitions

basic concepts:

symptoms, signs, test results, clinical findings

intermediate concepts:

pathophysiological

states

abstraction:rules, type-1 & type-2 fuzzy sets, temporal abstraction

feature extraction:

mean values, scores, …

preprocessing: missing data, plausibility, …

ICU patient data bases

y inference stepsreasoning

symbols

data-to-symbol

conversion

data

x inference steps

layer n-x-y-1

layer 2

layer 1

layer n-x-y

layer n-y

layer n (goal)

layer 0 (start)

… ……

Results

•

24 definitions of ICU-acquired infections –

6 definitions of bloodstream infections (BSI and BX)–

9 definitions of ICU-acquired pneumonias (PN)–

6 definitions of urinary tract infections (UTI)–

3 definitions of central venous catheter-related infections (CRI)•

Moni/Surveillance-ICU is operated at 12 ICUs at the Vienna General Hospital (96 beds); being extended to neonatal ICU patients, …

•

Moni/Surveillance-ICU is connected to HIS, LIS, and PDMS•

cockpit surveillance for infection control unit–

automated daily and/or manual activation•

Evaluation over a period of 2 months (2 ICUs)–

24 out of 28 patients TP (detected and correct), 0 FPs, 4 FNs

(technical reasons: missing data, missing in rule condition, …), many TNs

–

manual evaluation of criteria: each episode of infection > 2 hours–

with Moni: < 5 min per episode

Arden syntax

•

A standard language for writing situation-action rules that can trigger alerts based on abnormal clinical events detected by a clinical information system.

van Bemmel, J.H., Musen, M.A. (eds.) (1997) Handbook of Medical Informatics, Springer-

Verlag, Heidelberg, Glossary, p. 559.

•

A language to encode actions within a clinical protocol into a set of situation-action rules, for computer interpretation, and also to facilitate exchange between different institutions.

•

The Arden syntax resembles the Pascal computer programming language, and is procedural in its design.

Coiera, E. (2003) Guide to Health Informatics, Arnold, London, 2nd ed., p. 165.

Arden and Health Level Seven (HL7)

•

A standard language for writing situation-action rules that can trigger alerts based on abnormal clinical events detected by a clinical information system.

van Bemmel, J.H., Musen, M.A. (eds.) (1997) Handbook of Medical Informatics, Springer-Verlag, Heidelberg, Glossary, p. 559.

•

Each module, referred to as a Medical Logic Module (MLM), contains sufficient knowledge to make a single decision.extended by packages of MLMs for complex clinical decision support

•

Contraindication alerts, management suggestions, data interpretations, treatment protocols, and diagnosis scores are examples of the health knowledge that can be represented using MLMs.

extended by single and differential diagnosis, temporal monitoring, control systems, selective computerized processing of clinical pathways and management guidelines

•

The first version of the ARDEN syntax was administered and issued by the American Society for Testing and Materials ASTM (1992, version 1.0; today

2.5). Since 1998, an Arden Syntax Special Interest Group (SIG) is part of the HL7 organization (www.hl7.org).

functionality

integration

HIS, MIS, PDMS, LIS, medical practice SW, web-based EHR, telemedicine applications,health portals,…

reminders and alerts, monitoring, surveillance, diagnostic andtherapeutic decision support, …

*

* harmonized input data +collected reasoning data +knowledge application statistics

data services center

Arden, ArdenServer, and health care information systems

Moni/Surveillance-ICU and -NICU

•

identification and monitoring of nosocomial infections according

to KISS (German) and HELICS (European)–

47 MLMs–

data-to-symbol conversion done by MLMs–

fuzzy sets and operators simulated by MLMs

•

for NICU–

>100 MLMs

•

towards FuzzyArden

Contributions to success in CDSS

0% 50% 75% 100%

availability and significance of medical data, structuralization of

medical knowledge, standardization of medical work processes

knowledge

representationinference

mechanism

numbersstructuremedicine & health care processes

Some notes on the generality of medical knowledge

broadly-accepted medical knowledge

knowledge in MKPs

adaptation to

institution

knowledge options: selection; configuration; confirmation

data selection: preprocessing;

feature selection

personalization of medical knowledge

(sex, age, context)

individual patient at institution

ΔK

=given

ΔI →

as small as

possible(by standardization)