Siff Clinical Informatics Boards study set 1
Stages of Meaningful use
Stage 1: focused on capturing and sharing data - 2011 Stage 2: focus on advanced care processes with DSS - 2014 Stage 3: focus on improved outcomes - 2017
Attributes of a Learning Health Care system
adaption to the pace of change stronger syncrhony of efforts culture of shared responsibility new clinical research paradigm CDS universal electronic health records tools for database linkage, mining and use notion of clinical data as for the public good incentives aligned to evidence based practice public engagement trusted scientific broker leadership.
Basic Local Alignment Seach Tool (BLAST)
algorithm for determining the optiomal genetic sequence alignments based on observations that section of proteins are often conserved without gaps so using statistical analysis small subsequences can be used to narrow the search when looking for matching sequences in a database
the most effective way to reduce health care costs is:
bundled payments
Utility
can be used to determine the value of an outcome node Standard gamble - choose between X time in state of illness vs therapy with a known risk or cure rate Time trade off (TTO) choose between time X in state of illness vs Y time in perfect health If 4 years of perfect health = 10 years in current state then TTO = 0.4 and 3 years in current state then = 3 x 0.4 = 1.2 QALY
Arden Syntax for Medical Logic Module
coding scheme or language providing canonical means for writing rules relating to specific patient situations to appropriate actions for practitioners to follow. Maintained by HL7 puts local database / data needs in curly braces in the implementation of a medical logic module - these then need to be built locally reducing the portability of the MLM's
Blios' Funnel
cognitive breadth of diagnostic considerations are refined and restricted (narrowed) over the course of the interation between patient and doctor
Active Failures
errors that occur in an acute situation, the effects of which are immediately felt
Anchoring bias
failure to adjust the probability of a disease or outcome desipite being given new information = premature closure
The Fundamental Theorem of Biomedical Informatics by Freidman
"A person working in partnership with an information resource is 'better' than the same person unassisted" 3 corollaries 1. Informatics is more about people than technology 2. In order for the theorem to hold the resource must offer something the person does not already have - it must increment the persons knowledge in some way. 3. Whether the theorem holds depends on an interaction between the person and resource, the results of which cannot be predicted in advance.
What is clinical decision support?
A tool (may not be electronic) which reduces the cognitive burden on a user possibly providing structured guidance based on patient specific inputs
Argument
A word or phrase that helps complete the meaning of a predicate
SpIN
A specific test rules a condition in You can trust a positive test (because negatives are reliable so if its positive the pt probably really has the disease) Good for confirmation
Prevalence
# w a disease / population TP+ FN/TP+TN+FP+FN as prevalence of disease goes down the PPV goes down and number of FP goes up the higher the prevalence the more likely that people who test positive will actually have the disease Baseline prevalence = pretest probability in a table you would
Job Categories in informatics
*Academic* - informatics researcher or teacher *Professional* - CMIO, CNIO, Developer, Trainer *Liaison* - represent clinical group in IT initiatives, champion
ARPANET
*Advanced Research Projects Agency Network* A WAN created by the Defense Advanced Research Projects Agency (DARPA) of the DOD - became the precursor to the internet
ANSI
*American National Standards Institute* - private organization that oversees voluntary consensus standards
HITECH
*Health Information Technology for Economic and Clinical Health Act* Subtitle A: creates MU and puts ONC in charge of NwHIN Subtitle D: has HIPAA changes $27 Billion for MU workforce development
Affordable care act 3 goals
*Regulation and coverage* (remove coverage restrictions, high risk pools, no preexisting denial reasons, increase payroll taxes on upper incomes) *Major expansion of coverage* - individual mandate, exchanges, employer penalties, Medicaid expansions to 133% of poverty) *Bending the cost curve* - provides incentives for higher quality, lower cost care
3 categories of health care quality
*Structural* - factors that make it easier or harder to deliver high quality care (availability, staffing, etc) *Process* - factors describing health care content and activities (screening, following guidelines) *Outcomes* - Changes attributable to care (morbidity, mortality, functional status)
Common Rule
*The federal policy for the protection of human subjects* Published in 1991 after careful consideration of the Belmont proceedings Outlines basic provisions of IRB, informed consent and assurances of compliance
Overall MU requirements
- Use certified EHR technology in a meaningful manner -the certified EHR must provide for HIE to improve quality of care - submit information on clinical quality measures using the certified EHR
From data and information to knowledge
...
False negative rate
1-Sensitivity
False positive rate
1-Specificity
Key components of Clinical Decision Support
1. A knowledge base 2. Patient specific information 3. A mode of communication All lead to a CDS intervention
5 health care system goals of meaningful use
1. Improving safety, quality and efficiency 2. Engaging patients in their care 3. Increasing coordination of care 4. Improving the health status of the population 5. Ensuring privacy and security All MU criteria must map to one of these goals
Security Practices: Technical and Procedures to implement
1. Individual authentication of users 2. Access controls 3. Audit trails 4. Physical security and database recovery 5. Protection of remote access points 6. Protection of external communications 7. Software discipline 8. Software assessment
Security Practices: Organizational to implement
1. Security and confidentiality practices 2. security and confidentiality committees 3. Information security officers 4. Education and training programs 5. Sanctions 6. Improved authorization forms 7. Patient access to audit logs
5 Goals for the health system under MU - all MU criteria must map to one of these goals
1. improving safety, quality and efficiency 2. engaging patients in their care 3. increasing the coordination of care 4. improving the health status of the population 5. ensuring privacy and security
Rules for 21st century healthcare IOM
1. patient needs and values should drive variations in care 2. care should be based on continuous healing relationships (24/7) and by all modalities 3. patient as source of control 4. shared knowledge, free flow of information and transparency of information 5. anticipation of needs rather than reacting to them 6. evidence based decision making
Warner adage on the make up of informatics
10% Medicine 10% Technology 80% Sociology
SnOUT
A sensitive test rules a condition out You can trust a negative test Good for screening
Learning Health Care System from the IOM
A health system designed to generate and apply the best evidence for the collaborative health choices of each patient and provider; to drive the process of discovery as a natural outgrowth of patient care and to ensure innovation, quality, safety and value in health care It re-envisions the way patients are anchored into the system and looks for ways to do research with a real impact on patient care
Positive predictive value
Assume you know the test result and want to know the probability of the disease state is impacted by prevalence PPV = TP/TP + FP PPV = precision in information recall PPV = *precision* in information recall *Evoking strength* = PPV in a systems approach like INTERNIST1
APACHE III
Acute Physiology and Chronic Health Evaluation, version III
AHIMA
American Health Information Management Association
ARRA
American Recovery and Reinvestment Act of 2009
ASCII
American Standard Code for Information Interchange 7 bit code to represent alphanumeric characters
Stimulus bill
Another name for ARRA which contains HITECH (Health Information Technology for Economic and Clinical Health Act)
Core Functions of Public Health
Assessment Policy development Assurance (enforcement)
AMDIS
Association of Medical Directors of Information Systems
Negative predictive value
Assume you know the test result and want to know the probability of the disease state is impacted by prevalence NPV = TN/TN + FN
Triple Aim of Health Care Quality (from ACA)
Better Health Better health care Lower cost
Confidentiality
Confidentiality is about the information Information is not available or disclosed to unauthorized people and is released only in a controlled manner
Metcalfe's Law
Considers the value of a telecommunication system (ethernet) but originated in the consideration of older communicating devices (Fax) stating that the value of the network is proportional to the square of the number of connected user in the system: = n(n-1)/2
Meaningful Use requirements
See the tables in the notes Stage 1: EP 15 core and 5 of 10 menu EH 14 core and 5 of 10 menu one menu must be a public health measure Stage 2: EP 17 core and 3 of 6 menu EH 16 core and 3 of 6 menu
The Greek Oracle Problem
The finding that computers were not good at replacing people which led to the current efforts as outlined by Freidman that the goal of informatics is to augment people not replace them.
Curley Braces Problem
The issue in the arden syntax where the need for local logic / rules in a system are represented within curley braces. The problem is this must be built for each individual system and is rarely easily moveable between systems.
Alert Message
a computer generated warning which is activated when a record meets certain pre-specified criteria often indicating action is needed
New 2014 Clinical Quality Measures
No longer a core objective but still required to meet MU EP 9 of 64 CQM EH 16 of 29 must select CQM from 3 of the 6 NQS key health policy domains 1. patient and family engagement 2. patient safety 3. care coordination 4. population and public health 5. efficient use of heatlhcare resources 6. clinical processes / effectiveness must be sent electronically
Decision analysis tree
Decision node is a *square* Chance node is a *circle* and each branch is assigned a prob all branches at node must sum to P = 1 Outcome node is a *triangle* each is assigned a value if outcomes are life vs death then 1 and 0 are the values *Rollback analysis* - work your way back through the tree to compare the expected value at each branch of the decision
European Convention on Human Rights
Everyone has the right to respect for his private and family life, his home and correspondence. Public authorities should not interfere except as needed for public safety, the well being of the country, the prevention of crime and preservation of health or protection of the rights of others.
Fourth Amendment
From the Bill of Rights protects the safety of people in their homes, papers and effects and against unreasonable search and seizure
Universal Declaration of Human Rights
From the United Nations Article 12 No one shall be subjected to arbitrary interference with his privacy, family, home or correspondence, nor to attacks on his honor and reputation. Everyone has the right to the protection of law from such attacks.
Specificity
Given the patient does not have the disease wha tis the probability of a negative test. A characteristic of the test not impacted by prevalence SP = TN / TN+ FP minimizes false positives
Predictive value
Given the patient has a positive or negative test what is the probabilty the patient does / does not have the disease you know the result and are trying to determine the disease state
Sensitivity
Given the patient has the disease what is the probability of a positive test. Is a characteristic of the test. Not impacted by prevalence SN = TP/TP+FN = True positive rate Minimizes the false negatives called *recall* in information theory called *frequency* in systems approach like INTERNIST-1 "Recall that I'm frequently sensitive"
Screening test
Goal is to minimize False Negatives Want a high true positive rate So a highly sensitive test is good for screening
HIMSS
Health Care Information Management and Systems Society
IOM quality aims for 21st century healthcare
Health care should be: 1. *Safe* 2. *Effective* (avoid care unlikely to provide benefit) 3. *Patient centered* - respectful of patient preferences, needs and values 4. *Timely* 5. *Efficient* - avoid waste 6. *Equitable*
Independent Payment Advisory Board
IPAB - tasked with changing Medicare coverage and reimbursement when cost targets not met
Impact of prevalence on test interpretation
If know the SN and SP of the test FP rate is 1-SP TP is based on the prevalence x SN (the true positive rate) so as TP rate falls with decreasing prevalence the PPV falls
Aggregations
In information retrieval are collections of content from a variety of content types (bibliographic, full-text, annotated)
Annotated content
In information retrieval content that has been annotated to describe its type, subject matter and other attributes
Privacy
Is about the individual and their desire to limit the disclosure of personal information.
Clinical Informatics
Is the intersection between Information technology and Health Care.
Incremental Cost Effectiveness Ratio (ICER)
Is this treatment cost effective? Cost and utility / value of an outcome lets you compute this ICER is a measure of the change in cost with change in a unit of effectiveness (often QALY) ICER = (C1-C2)/(E1-E2) a negative value may indicate savings
HIPAA Changes in HITECH
Now applies to Business associates Civil and criminal penalties New breach notification standard - breach is presumed to impact individuals unless a covered entity does a risk assessment showing low risk - eliminated "risk of harm" standard Marketing of PHI limited where entity gets compensated Sale of PHI Research authorizations can look at future research as well as considering conditioned vs unconditioned issues Restrictions on disclosure if pt paid out of pocket Fundraising changes
Expected utility
a function of value but also of risk tolerance applies peronal preferences or issues to the base of expected value
Key Skills for CMIO's and priorities from Gartner study
Key skills - Leadership - Communication - Consensus building - Psychology, change management, finance/budgeting Priorities headed toward analytics and optimization from MU Need C level support and strategic commitment to IT from the organization
Automated indexing
Most common method of full text indexing. words are stripped of suffixes, entered as items in the index, then weights assigned based on ability to discriminate among documents
National Institute for Health and Clinical Excellence
NICE from the UK Uses QALY as part of cost effectiveness analysis to chose drug options
Asynchronous Transfer Mode (ATM)
Network protocol for sending streams of small fixed length cells of information over high speed dedicated lines
3 categories of clinical human research
Patient oriented where researcher interacts with patient Epidemiological and behavioral studies outcomes and health services research
Heuristic
a rule of thumb used in decision making
Belmont Report
Stressed 1. Respect for persons/autonomy their right to make choices based on personal values and beliefs 2. Justice - treat others equitably, fair distribution of benefits and burdens 3. Nonmaleficence - do no harm 4. Beneficence - do good Belmont report led to the common rule which led to IRB regulations
receiver operating characteristic
TPR (SN) vs FPR (1-SP) trade off between SN and SP shoulder or inflection point can help determine the optimal threshold for a test with a wide range of results Area under the curve is largest in the best test (optimizes TPR and minimizes FPR)
Health Care Stakeholders
The 5 P's 1. Patient (consumer, citizen) 2. Provider 3. Purchaser (govt, employers, individuals) 4. Payer - (insurance company, govt) 5. Public Health
Bayes Theorem in words
The relationship between the probabilities of A and B and the conditional probability of A given B and B given A.
Translational research
Trying to accelerate moving results of research from bench to bedside 3 axes 1. clinical efficacy - what care works 2. Outcomes research - who benefits from promising care 3. activities to test how to deliver high quality care reliably and in all settings
Code of fair information practice
US Dept of Health, Education and Welfare 1973 1. No personal data keeping systems which are secret 2. People must be able to find out what is in their record and how the information is used 3. People should be able to control the use of their data 4. There should be a way for people to correct or amend data about them. 5. Any entity holding or using data must assure the reliability of the data and make precautions to prevent misuse.
Confirmatory test
Want to minimize false positives Want a high true negative rate so a highly specific test is a good confirmatory test
Authorization
What a user is allowed to do in the system (based on job role, clearances, etc)
Quality Adjusted Life Year QALY
lets you model a wide range of conditions can be negative because some states of health may be considered worse than death how many years of perfect health are equal to 10 years in current state
Factors that make up security of systems
measures to protect information and systems including maintaining *confidentiality* of data as well as *integrity* and *availability* of the information and the systems used to access it.
"What If" or sensitivity analysis
model a range of probabilities what if the probability of treatment success due to med X is between 10 and 30% - at what threshold is the treatment too risky
Value induced bias
overestimate the probability of an outcome based on the value or risk associated with that outcome "it would be horrible to miss a brain tumor in the pt with a headache so better get a CT"
Representativesness bias
overestimating rare diseases by matching patients to the typical picture of that dz insenstive to pretest probability - a patient with a typical picture of a rare disease is still more likely to have an atypical presentation of a common disease When you hear hoof beats think horses not zebras
Availability bias
overestimating the probability of unusual events due to recent or memorable instances you are more likely to recall unusual instances to can mislead you "last patient I saw like this had disease Y so we should test for Y"
Artificial Neural network
program that performs classification by taking as input a set of findings that describe a given situation, propagating calculated weights through a network of several layers of interconnected nodes and generating output as a set of numbers where each output corresponds to the likelihood of a particular classification that could explain the findings
Authentication
proving you are who you say you are via passwords, biometrics, etc.
RHIA RHIT
registered health information administrator registered health information technician
Informatics definitions all..
revolve around the idea that its about the use of information not just the data itself.
Analog signal
signal that takes on a continuous range of values
Issues in US health care
spend alot of money for poor outcomes pay more for all aspects of care than other countries approaching 20% of GDP in health care spending
Patient Centered Outcomes Research Institute (PCORI)
supports comparative effectiveness research a public - private partnerships
Non-repudiation
the ability to ensure a party to a contract or communication cannot deny the authenticity of their signature or that they sent a document they originated (or that they did something in an EHR)
Backward chaining
used in rule based systems system looks at the left of a rule to determine if its is true by invoking other rules that can conclude the values of the variables in the rule of interest which are currently unknown and reference by the rule in quesiton done recursively until all rules that could supply the required values have been considered
Expected value
what you get if win vs loose if 1/80 chance of getting $1000 your expcted value is (1/80 x 1000) + (79/80 x 0) = 12.50
