Tuesday, October 11, 2011

Classifying Adverse Events

I'm continuing my review of the NIST Technical Evaluation, Testing and Validation of the Usability of Electronic Health Records (NISTIR 7804).  Today I'm going to examine how they classify adverse events starting on page 21, because I see several weaknesses in the classification system they've prevented.

Adverse Events are one of the four main components of the model for analysis and understanding of use related risks of EHR systems found on page 18 and reproduced below:
  1. Use Error Root Causes—Aspects of the user interface design that induce use errors when interacting with the system.
  2. Risk Parameters—These are attributes regarding particular use errors, i.e., their severity, frequency, ability to be detected, and complexity.
  3. Evaluative Indicators—Indications that users are having problems with the system that are identified in direct observations of system use by the evaluation team in the environment of use or interviews with users.
  4. Adverse Events—A description of the outcome of the use error, and standard classification of patient harm.
The diagram shows the classification of patient harm, for which there are a variety of good classification systems that fit the degree of patient harm shown in the figure above.  The description of the outcome of the use error for Adverse events is listed below:
Wrong patient action of commission:
Actions with potentially harmful consequences are performed for one patient that were intended for another patient primarily due to inadequate selection mechanisms or displays of patient identifiers,
Wrong patient action of omission:
A patient is not informed of the need for treatment primarily due to inadequate selection mechanisms or displays of patient identifiers,
Wrong treatment action of commission:
Treatments that were not intended for a patient are provided primarily due to inadequate selection mechanisms or displays of treatment options,
Wrong treatment action of omission:
Treatments that were intended for a patient are not provided primarily because of inadequate selection mechanisms or displays of patient identifiers,
Wrong medication:
A patient receives the wrong medication type, dose, or route primarily due to inadequate selection mechanisms or displays of medication data,
Delay of treatment:
A patient receives a significant delay in the provision of care activities due to design decisions made to satisfy billing, security, or quality improvement objectives, and
Unintended or improper treatment:
A patient receives unintended care due to misunderstanding how to provide care in the system or due to actions taken to test software, train users, or demonstrate software to potential customers.
One of the challenges of this list is that it includes causes following the "due to ..."
Another problem is that I have difficulty understanding how to classify something such as "Patient was given the wrong dose of an immunization".  I could classify it as "wrong treatment/commission", "wrong medication", or "impromper treatment".  The classification vocabulary is poorly defined.  From the principles for creating a classification system, it doesn't cover the classification space, and each item in it doesn't stand alone (and no hierarchy is assigned to address overlaps).

There are a number of good papers on how to classify adverse events.  They suggest several axes for classification:
  1. The JCAHO Patient Safety Event Taxonomy: a standardized terminology and classification schema for near misses and adverse events.
    This report classifies adverse events by four axes:  Impact, type, domain and cause, and discusses another axis for prevention.
  2. WHO Draft Guidelines for Adverse Event Reporting and Learning Systems.
    See page 19 on Classification of Events (Adverse Event, Error, Near Miss/Close Call, Hazard/Unsafe Condition), Page 23 for a brief discussion on type of event, Page 24 for Probability and Severity classifications, 
  3. Classification of Medical Errors and Preventable Adverse Events in Primary Care: A synthesis of the literature.
    This report classifies by "Type of Event" and "Process Factors"
The one that all seem to agree upon is "type of event", which describes "what went wrong".  From a clinical perspective, it seems that the highest level of what went wrong is inappropriate treatment, diagnosis or preventative services.  This can be further sub-classified as incorrect, contraindicated, incomplete, delayed or no service provided.  With medications it can be further sub-classified as wrong medication, dose, route or frequency.  Wrong route in medications corresponds to wrong site in procedure.

Note:  Wrong patient is a very commonly associated with adverse events, but doesn't fit into the classification of type of event, it fits into cause.

I think that NIST should be looking at a very strong taxonomy of adverse events, such as that described by the first report by JCAHO listed above.


Post a Comment