Going paperless: Going paperless: from drowning in paper from drowning in paper to drowning in data to drowning in data Dr Jeremy Rogers MRCGP Medical Informatics Group Department of Computer Science University of Manchester www.cs.man.ac.uk/mig i MedicalInformaticsGroup is presentation has extensive speaker’s not
26
Embed
Going paperless: from drowning in paper to drowning in data Dr Jeremy Rogers MRCGP Medical Informatics Group Department of Computer Science University.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Going paperless: Going paperless: from drowning in paper to from drowning in paper to
Medical Informatics in the UKMedical Informatics in the UK
General Medical Council: Tomorrow’s Doctors (1993)
One area of particularly rapid expansion has been in the application of computers to medicine. The extent to which future advances may revolutionise not only systems of communication, but also care procedures and possibly education itself, is unpredictable, but a working knowledge of modern medical information technology will be essential for the doctor of the future.
At the end of the course of undergraduate education the student will have acquired … the other essential skills of medicine, including
(a) basic clinical method (history & examination)(b) basic clinical procedures (life support, venepuncture)(c) basic computing skills as applied to medicine
Why go paperless -Why go paperless - The medical malaise The medical malaise
(with thanks to Larry Weed)
Medical Malaise: The SymptomsMedical Malaise: The Symptoms
• Medical Knowledge growing ever faster
• Super-specialisation– focus on ever smaller parts of the problem
• ‘collusion of anonymity’ (Balint 1964)
• stereotypical diagnostic and treatment behaviour
• holistic approaches regarded with suspicion
• Meanwhile..– patients are not a collection of discrete problems
– patients becoming expert on their own condition
Medical Malaise: The SymptomsMedical Malaise: The Symptoms
• Huge variations in outcome and cost– standardised mortality rates (119 in Walsall cf 68 at UCLH)
– cancer survival rates– more patients killed annually in US through medical accident than on
roads [IOM Kohn, Corrigan et al. 2000]
– 11% of UK hospital patients harmed by medical error, half were avoidable, one third serious or fatal [BMJ 2001;322:517-519 ( 3 March )]
• Empirical observation: evidence for best practice doesn’t translate into altered practice
• NHS funding may have masked issue in UK– but won’t for much longer
– phenomenon just as obvious in well funded systems
Medical Malaise: The DiagnosisMedical Malaise: The Diagnosis
Information overload:
Drowning in Paper
‘The scarcely tolerable burden of information that is imposed taxes the memory but not the intellect’
(GMC 1993)
Medical Malaise:Medical Malaise:Patient is in denialPatient is in denial
• Pretence that we can cope– ‘Trust me, I’m a doctor’
• Refuge in good intentions– OK, so I’m not infallible, but I mean well
• Defence of clinical freedom– To do what, exactly ?
Medical Malaise: Medical Malaise: Patient is in denialPatient is in denial
• Shipman, Bristol: simple stats could have detected– Don’t trust me, I’m a doctor
• Clinical Audit– Death isn’t the only poor outcome we could measure
• is it the only one we have a language for ?
– Do I really want to know just how fallible I am ?• How could I possibly do better ?
• League Tables– Better not frighten the horses (in case they trample me)
Options for Treatment:Options for Treatment:Tighten Manual SystemsTighten Manual Systems
• Disseminate evidence– NICE, CHI
• Regularise Practice– Health Improvement Plans– Careplan pathways– National Service Frameworks– Formularies and protocols
• Monitor performance– GMC– Clinical Audit
Options for Treatment:Options for Treatment:Stop pretendingStop pretending
‘The burden of factual information imposed on students in undergraduate medical curricula should be substantially reduced’ (GMC 1993)
Q: who (or what) takes up the burden of knowing the facts ?
Options for Treatment:Options for Treatment:Information TechnologyInformation Technology
• The Computer Based Record – digitised (typed, dictated, OCR, images)– info can be in more than one place at a time– relatively easy to implement
• mostly a plumbing and data storage issue
• Computer Based Knowledge Repositories– NELH, PHLS
The Computer Based EPR:The Computer Based EPR:How to Drown in DataHow to Drown in Data
• Computers can not read !– find me all patients with joint disease
– which protocol should this patient be on ?
– is this patient on an anti-anginal ?
– is this cryptosporidium part of a known outbreak ?
– how to link record to knowledge repository ?
• Electronic ‘fat folder’ worse than physical one• Computer as passive conduit: GIGO
Drowning In DataDrowning In DataEPR - Dr Kildare - 26th Oct 2000
John Doe36 yrsEngineerMarried, 2 children
12.10.96 GP Surgery: Dr Kildare13.10.96 GP Surgery: Dr Kildare 20.10.96 GP Surgery: Dr Finlay 24.10.96 GP Surgery: Dr Kildare 10.11.96 GP Surgery: Dr Kildare 12.11.96 Radiology: reported film 27.11.96 GP Surgery: Dr Kildare 07.03.97 GP Surgery: Dr Kildare 19.04.97 GP Surgery: Dr Kildare 01.06.97 GP Surgery: Dr Kildare 18.10.97 GP Surgery: GP Registrar 03.03.98 GP Surgery: Dr Kildare 04.03.98 Path Links: WCC result30.06.98 GP Surgery: Dr Kildare 15.09.98 GP Surgery: Dr Kildare 05.11.98 GP Surgery: GP Registrar 03.01.99 GP Surgery: Dr Kildare 17.02.99 GP Surgery: Nurse Duffy21.03.99 GP Surgery: Dr Kildare 07.10.99 GP Surgery: GP Registrar 26.01.00 GP Surgery: Nurse Duffy
Encounters
Active ProblemsAsthma
Current MedicationSalbutamolHydrocortisone
PEFRBP WCC
Alerts / RemindersAsthma checkBPFlu Vaccine
This VisitCode Notes ActionPEFR 550 l /min
C/o Low Mood Declined antidepressantAsthma Influvac im BN #035679A4
Salbutamol inh 2 puff qds 1opChest NAD. No Problems.
Letters Results Appt
Drowning In DataDrowning In DataEPR - Dr Kildare - 26th Oct 2000
C/o Low Mood Declined antidepressantAsthma Influvac im BN #035679A4
Salbutamol inh 2 puff qds 1opChest NAD. No Problems.
Letters Results Appt
Drowning In Data:Drowning In Data:Data analysis and Coding ChaosData analysis and Coding Chaos
Sore Throat Symptom 0.6 117
Visual Accuity 0.4 644
ECG General 2.2 300
Ovary/Broad Ligament Op 7.8 809
Specific Viral Infections 1.4 556
Alcohol Consumption 0 106
H/O Resp Disease 0 26
Full Blood Count 0 838
How How notnot to drown in data: to drown in data:1. Ask what the record is for1. Ask what the record is for
• Post-hoc documentation for medico-legal protection ?• Aide-memoir of treatment plan for author ?• Aide-memoir of treatment plan for team ?• Objective record of patient state ?• Input for decision support ?• Part of population data-set
– for resource planning ?
– for care quality analysis ?
– for data mining ? ?
How How notnot to drown in data: to drown in data: 2. Examine The Typical Medical Record2. Examine The Typical Medical Record
• Much is only implied• Ambiguity and imprecision is rife
– gastrointestinal disturbance: what does this mean ?
• Significant findings often not recorded– focus on significant positive findings is usually a good strategy– except when it isn’t– symptomatic of traditional diagnostic process
• common things are common• rush to stereotype patients
• Problem, Plan and checkpoints not stated or linked
How How notnot to drown in data: to drown in data: 3. Conclude: current MR inadequate3. Conclude: current MR inadequate
• Written MR does not support all tasks even when you attempt to perform them manually– therefore computers have no chance
• Need to tighten up EPR capture, and tune it to intended tasks– need much more explicit info
– EPR as active resource, not historical document
– Input and Output must be developed together
Options for Treatment:Options for Treatment:Computer as active agentComputer as active agent
• Computerised EPR– digitised (but not typed, dictated, OCR, images)
• because computers can not read
– info can be in more than one place at a time
– computer must also ‘understand’ content• recommend protocols• measure quality of care• find patient groups• filter record• data mine
– otherwise it can’t help
Why its hardWhy its hard
• can’t read• have no context to resolve
ambiguity• need complete data• require precision• are capable of coherent
reasoning too complex for a human to comprehend (or debug, or trust)
• never tire
• prefer natural language
• are often non-specific
• are rarely complete
• prefer vagueness
• think heuristically
• are easily bored
Computers Humans
Why its urgent: Why its urgent: the new medicinethe new medicine
• The new genetics– idiosyncratic patient responses are genetically determined– mechanisms not understood (or, even, understandable)
• Solution:– data mine to associate genotype with phenotype
• Problem:– how to describe phenotype consistently and completely
• when you don’t know what you’re looking for
– patient record becomes primary diagnostic tool
ConclusionConclusion
• Already drowning in paper– but reluctant to admit it
• Risk drowning in data– computer based EPR not a solution
– need computer as active partner, not passive conduit
• Information explosion imminent– either we must recruit computers
– or we choose not to use the information• but people will suffer